Re: [Emc-developers] Machinekit slowdown ?
On 6/3/24 11:56, Jon Elson wrote: I have several special-purpose devices based on machinekit on the Beagle Bone. In one case, we are using it in a hospital where they won't allow us to connect anything to their network. So, we set up a point to point Ethernet link between a laptop and the Beagle Bone. When starting LinuxCNC, there is a HUGE delay of about 10 minutes before the Axis GUI comes up. I think I have traced it to be caused by there being no NTP time server. I think we see the same delay with either a Linux or Windows computer used to access the Beagle. Does this ring a bell with anyone? Is there a workaround? (Process to kill or needs to be run?) Thanks for any hints, Jon There are two solutions Jon. The latest crop of ntp offerings can be both client and server. ntpsec is one of them, and more difficult to setup but I have it running on this machine as both client and server, so I am a stratum 3 server to the rest of my stuff. Means I'm within a millisecond of the clock in Co Springs. Chrony can also do that and may be easier to setup. Its relatively easy to do so I'm not banging on debians pool of servers except for this machine. Using chrony on all my pi clone stuff. Bypassing it can I'm sure be done, but the proper thing would be for the hospital to setup its own ntp server and allow your machine access since there are things that cannot be well done if the system clock is wrong. I'd not be bashful about presenting that argument to the hospitals net controller. Its not that hard, even for a winders system, to cut a single ipv4 address into their system. You could even restrict that address to only the ntp port #. It should of course be a non-routable address, someething in the 192.168.xxx.yyy range. Any solution I would offer demands a net access. It is important enough that its non-negotiable. Good luck ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] vismach example movies
On 5/23/24 14:00, Stuart Stevenson wrote: https:/l/videobin.org/+87f/b08.html https://videobin.org/+87g/b09.html https://videobin.org/+87i/b0b.html https://videobin.org/+87j/b0c.html https://videobin.org/+87k/b0d.html These are all the links on the wiki page. heh - I copied them all but i suspect if one works then all will work thanks Stuart Here, Stuart, none of them work, service unavailable ack bookworm version of FF, but not upodate. I'll try again when synaptic is done. On Thu, May 23, 2024 at 11:20 AM andy pugh wrote: On Wed, 22 May 2024 at 20:14, Stuart Stevenson wrote: Gentlemen, The wikiQuick links to vismach movies links do not work. Which link? -- atp "A motorcycle is a bicycle with a pandemonium attachment and is designed for the especial use of mechanical geniuses, daredevils and lunatics." — George Fitch, Atlanta Constitution Newspaper, 1912 ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] The Wiki Certificate might have expired
On 5/17/24 18:27, andy pugh wrote: My Mac is really unwilling to visit wiki.linuxcnc.org Works as expected here Andy. linux, bookworm, latest FF. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] How closely do we need to adhere to the NISTstandard?
On 4/9/24 15:26, Andy Pugh wrote: On 9 Apr 2024, at 19:43, Nicklas SB Karlsson wrote: There might be requirements for safety reasons but do not think we need to care about some standard require MDI commands to be executed only with a physical button. This is actually already the case with Touchy, FWiW. And is one of the reasons I use it. I would expect that the removal of the return key, leaving only the go button to exec the mdi would not be all that difficult in axis. But I don't see a reason to change it. Perhaps another option to be configured in the .ini file. Has anyone actually objected to the present method? Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] How closely do we need to adhere to the NISTstandard?
On 4/9/24 09:02, andy pugh wrote: Apart from anything else, I am not sure that it is a standard. This is based on bug report #2956. Our current behaviour does not match the NIST RS274 interpreter "report" from 2000. The last actual "RS" standard (ie the RS in RS274) was issued in the 1970s. There is an ISO standard from 1982, and then the NIST EMC report from 2000. My own feeling is that any deliberate deviation from the NIST v3 report is not a "bug" as such, and that perhaps we should simply close the "bug" as a "feature". Has anyone else looked at this bug report? What do you think? https://github.com/LinuxCNC/linuxcnc/issues/2956 Currently we can click on a history line, re-use it or edit it before clicking on go or hitting enter. We are used to it. Please do not change it if the change disables our ability to edit before reuse. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] IGBT Module Dial-a-Yield Project Ideas
On 4/5/24 18:02, Curtis Dutton wrote: Gene can you send some links to some of these drives/motors. I'd like to read more about them. Sounds like interesting tech. Maybe we can steal some ideas here. Plus if we get a control board set up and running there is no reason we couldn't adapt it to run other types of power modules. I don't have any book marked. Best is to go to the hanpose web site and search for "closed loop steppers". Or for 3 phase stuff, google for LCDA357H. 3 phase only turns 1.2 degrees a full step but this is only a 50 volt driver I'm using with 42 volts, 90 volt nema 17 or nema 23-34 are LC42 or LC57, both of those can use psu's to 90 volts or more. On Fri, Apr 5, 2024, 12:44 PM gene heskett wrote: On 4/5/24 08:25, Curtis Dutton wrote: yessir thank you. I studied it a fair bit a few years ago. It's part of the inspiration for this for sure. I'd like this new board to run out of thr box linuxcnc hal and maybe we can port and adopt lots of the STMBL code into this project. My goal is to have this "controller" be a bone stock hal based system much like STMBL except not the "stripped" down hal that Rene created. Which I'd assume he had to do because of processing and memory power constraintd. I havent yet started playing with my pine star board but it should have more than enough power. Hopefully we can get at least a 4khz thread going on it. Maybe more! What are modern servo amplifiers running their pid loop at these days? Stepper/servo's, since two timers cost money, I suspect are running at their basic ultrasonic speed of 22khz or better. It simply does not make economic sense to do otherwise. On Fri, Apr 5, 2024, 8:13 AM andy pugh wrote: On Fri, 5 Apr 2024 at 11:59, Curtis Dutton wrote: I'm starting a new project to build a motor drive based upon igbt modules. Hoping to build something that has a single type of controller that will plug and play with various capacity igbt modules for various size motors. Are you aware of the STMBL project? That would seem like a very good starting point, as it already plays well with LinuxCNC, and works with a wide range of feedback types (Hall sensors, resolvers, encoders and serial protocols) In fact you could probably copy the LV board with only minor changes (to account for component obsolescence) and make a new HV/power board. https://github.com/rene-dev/stmbl/blob/master/README.md -- atp "A motorcycle is a bicycle with a pandemonium attachment and is designed for the especial use of mechanical geniuses, daredevils and lunatics." — George Fitch, Atlanta Constitution Newspaper, 1912 ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] IGBT Module Dial-a-Yield Project Ideas
On 4/5/24 08:25, Curtis Dutton wrote: yessir thank you. I studied it a fair bit a few years ago. It's part of the inspiration for this for sure. I'd like this new board to run out of thr box linuxcnc hal and maybe we can port and adopt lots of the STMBL code into this project. My goal is to have this "controller" be a bone stock hal based system much like STMBL except not the "stripped" down hal that Rene created. Which I'd assume he had to do because of processing and memory power constraintd. I havent yet started playing with my pine star board but it should have more than enough power. Hopefully we can get at least a 4khz thread going on it. Maybe more! What are modern servo amplifiers running their pid loop at these days? Stepper/servo's, since two timers cost money, I suspect are running at their basic ultrasonic speed of 22khz or better. It simply does not make economic sense to do otherwise. On Fri, Apr 5, 2024, 8:13 AM andy pugh wrote: On Fri, 5 Apr 2024 at 11:59, Curtis Dutton wrote: I'm starting a new project to build a motor drive based upon igbt modules. Hoping to build something that has a single type of controller that will plug and play with various capacity igbt modules for various size motors. Are you aware of the STMBL project? That would seem like a very good starting point, as it already plays well with LinuxCNC, and works with a wide range of feedback types (Hall sensors, resolvers, encoders and serial protocols) In fact you could probably copy the LV board with only minor changes (to account for component obsolescence) and make a new HV/power board. https://github.com/rene-dev/stmbl/blob/master/README.md -- atp "A motorcycle is a bicycle with a pandemonium attachment and is designed for the especial use of mechanical geniuses, daredevils and lunatics." — George Fitch, Atlanta Constitution Newspaper, 1912 ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] IGBT Module Dial-a-Yield Project Ideas
ny speed above the speed of the internal A/D's. which on the present offerings, several milliseconds per step. They would be great in an emco5 tool changer clone. They are also limited to about 30 volt supplies so available accels suffer. Lesson is: optical encoders only for smooth motion. The higher resolution of the encoder, the more accuracy potential there is. For a discussion of this error, see a good text on Nuiquist error. Actually, I am hoping to see an rvs30 worm drive clone with adjustable backlash, so we could comp for wear-in. That would bring arc/second accuracy to my A & B axis's. Say hello to Better gear hobbing than Boston Gear. I know this has been discussed before and it's time to get this idea going. I'm tired of trying to source drives and vfds and having to learn how to operate, integrate and maintain all these different devices, when they are all functionally identical. (not to mention thay they are opaque and closed source) I'm starting out part of my investigation using a pine64 board. It's risc-v based soc with 2 ethernet ports, wifi, bluetooth etc... I'm going to initially interface it as an ethercat "device" My first step is to get linux cnc running on it and go from there. Does anyone have experience with anything risc-v related? Tips, advice, resources pretty much any pointers are welcome to help me get up to speed faster. Also I need an initial igbt module "family" to target. Does anyone have any suggesstions along those lines as well? Hope all is well. Thanks, Curtis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers . Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Status of LinuxCNC in Debian
On 3/22/24 06:13, andy pugh wrote: Is LinuxCNC currently in Debian? I know that we were due for removal from Testing, and since then I have heard no further updates from Debian about anything at all. IDK about testing/sid, I'm on stable and 2.9.0 is still there. personally our inability to get fixes needed into user space for users via being in a distro has turned into a PITA. I no longer see it as a feather in our collective hat. I see it today as wasting time over minutia although the docs are now much improved. My $.02. I did get my holes drilled. And in drilling them disclosed that the 2 linear bearings I chose out of the 4 I ordered, were not made on the same machine, I chose to use every third bolthole, which on 1 was about 59.98 mm while the other was 60.000 mm. I don't have accurate enough gauges to say which was correct. Or if either. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] This run once problem just won't to away.
On 3/13/24 15:19, gene heskett wrote: On 3/12/24 16:35, gene heskett wrote: On 3/12/24 13:49, gene heskett wrote: On 3/11/24 12:31, andy pugh wrote: On Mon, 11 Mar 2024 at 00:11, gene heskett wrote: Since no one has commented I updated to todays build, no help there. lcnc refuses to move the machine by my code more than once per startup. It sounds like the code leaves a parameter in a state that causes the code to skip. If it isn't error-ing then it is probably running to completion, but not doing what you expect. Try running through it in single-block mode. I got tired of looking at all the screaming qt was doing trying to get my camera to work, so I commented it out. Another and2 fixed the false atspeed indicator saying its at speed when stopped. But what I set for MIN_VELOCITY in the spindle_0 section of the ini has no effect. I still have to click the + button to start the spindle at about 105 revs. So now its warmed up, I'll go play. Cheers, Gene Heskett, CET. More of the same, might be a little more stable w/o all the qt screaming. It will Single Step to the end of the first hole while I fine tune some of the presets for the first hole, And as the center of x or y is found. I mam updating the 3 vars that the hole finder routine uses to recenter before starting the next measurement. But my run once per restart of linuxcnc continues. It will not restart the spindle on the 2nd try. This "first move" is in the o sub. The s<_searck_revs> command is a few lines prior to the M4, works the first time every time, very occasionally works 2nd time. When I know it should have spun up the spindle to around 1000 revs to find the linear rails top, the spindle hasn't turned and the atspd led is still red. No debug statements elsewhere in the code ever exec. No response at all to the step button but the stop button works fine. All indications are that atspeed is the culprit. I can debug the #<_search_revs> var, and it reads just fine. But an M4 on the mdi line does zip. I sure could use a clue. Latest, probably messy code attached. Thanks Andy. Take care & stay well. I found it, I think. When it fails, the speed signal value is being fed to pid.s.commamd, but pid.s.enable isn't being switched on, and I've located it in my hal, but I'm not visualizing it well since its nearly 10 yo code I've not modded since. Tracing backwards I'm getting lost in how I check the boot module for the pico-pwm-servo for an output of 5 at the state pin. I originally did a servo reboot at every spindle start. Its possible that I have lost the trigger signal in my mods to do the spindle turnaround as quick as I am. The idea is to delay the pid's output until the servo had been booted. I'll find it but its going to take more coffee. Which I've got to make. So call the St Bernards back in, its not LCNC's fault. It's mine and I'll own it. Thanks Andy. Cheers, Gene Heskett, CET. No, i won't own it. Someplace back in the fog of time, the comp module effectively got its in0 and in1 swapped so the man page is /now/ correct. It wasn't a decade back and I had to invert the .out to make it work then, now its correct and I no longer need the not if I swapped inputs. So I mow get an atspd indication that tells the truth. I s/b be fixed. Thanks Andy. Take care and stay well. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] This run once problem just won't to away.
On 3/12/24 16:35, gene heskett wrote: On 3/12/24 13:49, gene heskett wrote: On 3/11/24 12:31, andy pugh wrote: On Mon, 11 Mar 2024 at 00:11, gene heskett wrote: Since no one has commented I updated to todays build, no help there. lcnc refuses to move the machine by my code more than once per startup. It sounds like the code leaves a parameter in a state that causes the code to skip. If it isn't error-ing then it is probably running to completion, but not doing what you expect. Try running through it in single-block mode. I got tired of looking at all the screaming qt was doing trying to get my camera to work, so I commented it out. Another and2 fixed the false atspeed indicator saying its at speed when stopped. But what I set for MIN_VELOCITY in the spindle_0 section of the ini has no effect. I still have to click the + button to start the spindle at about 105 revs. So now its warmed up, I'll go play. Cheers, Gene Heskett, CET. More of the same, might be a little more stable w/o all the qt screaming. It will Single Step to the end of the first hole while I fine tune some of the presets for the first hole, And as the center of x or y is found. I mam updating the 3 vars that the hole finder routine uses to recenter before starting the next measurement. But my run once per restart of linuxcnc continues. It will not restart the spindle on the 2nd try. This "first move" is in the o sub. The s<_searck_revs> command is a few lines prior to the M4, works the first time every time, very occasionally works 2nd time. When I know it should have spun up the spindle to around 1000 revs to find the linear rails top, the spindle hasn't turned and the atspd led is still red. No debug statements elsewhere in the code ever exec. No response at all to the step button but the stop button works fine. All indications are that atspeed is the culprit. I can debug the #<_search_revs> var, and it reads just fine. But an M4 on the mdi line does zip. I sure could use a clue. Latest, probably messy code attached. Thanks Andy. Take care & stay well. I found it, I think. When it fails, the speed signal value is being fed to pid.s.commamd, but pid.s.enable isn't being switched on, and I've located it in my hal, but I'm not visualizing it well since its nearly 10 yo code I've not modded since. Tracing backwards I'm getting lost in how I check the boot module for the pico-pwm-servo for an output of 5 at the state pin. I originally did a servo reboot at every spindle start. Its possible that I have lost the trigger signal in my mods to do the spindle turnaround as quick as I am. The idea is to delay the pid's output until the servo had been booted. I'll find it but its going to take more coffee. Which I've got to make. So call the St Bernards back in, its not LCNC's fault. It's mine and I'll own it. Thanks Andy. Cheers, Gene Heskett, CET. ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] This run once problem just won't to away.
On 3/12/24 13:49, gene heskett wrote: On 3/11/24 12:31, andy pugh wrote: On Mon, 11 Mar 2024 at 00:11, gene heskett wrote: Since no one has commented I updated to todays build, no help there. lcnc refuses to move the machine by my code more than once per startup. It sounds like the code leaves a parameter in a state that causes the code to skip. If it isn't error-ing then it is probably running to completion, but not doing what you expect. Try running through it in single-block mode. I got tired of looking at all the screaming qt was doing trying to get my camera to work, so I commented it out. Another and2 fixed the false atspeed indicator saying its at speed when stopped. But what I set for MIN_VELOCITY in the spindle_0 section of the ini has no effect. I still have to click the + button to start the spindle at about 105 revs. So now its warmed up, I'll go play. Cheers, Gene Heskett, CET. More of the same, might be a little more stable w/o all the qt screaming. It will Single Step to the end of the first hole while I fine tune some of the presets for the first hole, And as the center of x or y is found. I mam updating the 3 vars that the hole finder routine uses to recenter before starting the next measurement. But my run once per restart of linuxcnc continues. It will not restart the spindle on the 2nd try. This "first move" is in the o sub. The s<_searck_revs> command is a few lines prior to the M4, works the first time every time, very occasionally works 2nd time. When I know it should have spun up the spindle to around 1000 revs to find the linear rails top, the spindle hasn't turned and the atspd led is still red. No debug statements elsewhere in the code ever exec. No response at all to the step button but the stop button works fine. All indications are that atspeed is the culprit. I can debug the #<_search_revs> var, and it reads just fine. But an M4 on the mdi line does zip. I sure could use a clue. Latest, probably messy code attached. Thanks Andy. Take care & stay well. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ( linear_bearings.ngc, copyright 25 Feb 2024 by Maurice E Heskett, CET ) #<_measure>=1. ( zero to drill after changing fr probe to 3mm cutter ) #<_hole_spc> = 60. ( nominal space beteen holes drilled ) #<_xy_spd> = 3 #<_z_spd> = 15 #<_search_revs> = 1500 (in M4 reverse) G17 G21 G40 G49 G54 G61.1 G80 G90 G94 ( setup to use xy arcs, metric, exact path) ( cancel diameter comp,cancel length off set, G54 coordinates system 1, ) ( cancel any canned cycles, absolute distances, feed per minute mode ) ( setup contact to ground to bearing, put probe in spindle with probe wire ) ( NOTE Andy Pugh says this is a valid way to address an array: #[100 + #3] ) ( but he didn't mention that debug was broken, so if we want to see it ) ( to a global var, we must first copy the #<_gvar> = #[basenum+#50] ) ( rather cumbersome proceedure ) (pre-alloocate the bounce vars for floats) (allocate vars, 6 point holding tank) #<_zx> = 0. (for z stuffs) #<_zy> = 0. (for z stuffs) #<_zz> = 0. (for z stuffs) #<_xyx> = 0. (for hole stuff) #<_xyy> = 0. (for hole stuff) #<_xyz> = 0. (for hole stuff) #50=0 ( set index to zeroth element then set starting points ) #[200+#50] = 2.6 ( zeroth x location for z_check ) #[210+#50] = 2.1974( zeroth y location for z_check ) #[220+#50] = -84 ( zeroth height to begin z search ) #[230+#50] = 11.500( zeroth hole xy x location ) #[240+#50] = 1.8424 (zeroth holes xy y location ) #[250+#50] = [-1.75 + #[220+#50]] ( to see if math works place xy probe 1.75mm into hole) (debug, init_arrays done) o sub ( now, copy all that to global vars ) #<_zx> = #[200+#50] (z_check x) #<_zy> = #[210+#50] (z_check y) #<_zz> = #[220+#50] (z_check z) #<_xyx> = #[230+#50] (hole_check x) #<_xyy> = #[240+#50] (hole_check y) #<_xyz> = #[250+#50] (hole_check z) ( preset 3 vars for hole_finder) #100= #<_xyx> #101= #<_xyy> #102= #<_xyz> o endsub o sub (debug,lb sa 47 hole=#50) (debug,lb sa 48 zx=#<_zx> #<_zy> #<_zz>) (debug,lb sa 49 xyx=#<_xyx> #<_xyy> #<_xyz>) (debug,lb sa 50 #100 #101 #102) o endsub o sub ( uses hole arrays 230-240-250 but update all for this hole ) #[230+#50] = #100 (metric x return from hole_finder) #[240+#50] = #101 (metric y return from tholefinder) #[250+#50] = #102 (metric z return from latest g38.2 z-75) o endsub o sub ( last command of main loop ) ( will make
Re: [Emc-developers] This run once problem just won't to away.
On 3/11/24 12:31, andy pugh wrote: On Mon, 11 Mar 2024 at 00:11, gene heskett wrote: Since no one has commented I updated to todays build, no help there. lcnc refuses to move the machine by my code more than once per startup. It sounds like the code leaves a parameter in a state that causes the code to skip. If it isn't error-ing then it is probably running to completion, but not doing what you expect. Try running through it in single-block mode. I got tired of looking at all the screaming qt was doing trying to get my camera to work, so I commented it out. Another and2 fixed the false atspeed indicator saying its at speed when stopped. But what I set for MIN_VELOCITY in the spindle_0 section of the ini has no effect. I still have to click the + button to start the spindle at about 105 revs. So now its warmed up, I'll go play. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] This run once problem just won't to away.
On 3/9/24 20:38, gene heskett wrote: This gcode, when I can get it to run, finda all 8 holes just fine. And now it will not run twice, refusing to go past the first move command on the 2nd try after a full restart, and rehome.. With no reported errors on the first run.. This is the current copy attached. nuke any others you might have before saving it, I won't touch till someone tells me to. Thanks all. Take care & stay well. Cheers, Gene Heskett, CET. Since no one has commented I updated to todays build, no help there. lcnc refuses to move the machine by my code more than once per startup. Does not highlight the failing line, no error messages except related to my camera which doesn't work on the terminal its started from. debug statements work up to the first s1000 call, but no further, the machine CAN move by hand or be rehomed but my code can't even start the spindle. It will do anything I've tried from the mdi. The only thing different between this code and other code I've written is the use of #[2xx+#50] syntax, where #50 is the loop counter start it 0, max 7, to access 6 10 element arrays, starting at 200,210,220,230,240 & 250, theoretically using 8 of the 10 locations which the part I have written uses to contain the xyz locations of the locations about 5mm away from each hole for the Z check location , that is 3 of the 6 arrays, and 3 more arrays to contain the XY's found for each of the 8 holes. Seems pretty simple and conventional. But w/o any error msgs, indicating what died, how should I proceed? The backplot has no problems displaying it including my mistakes. The coupling between the backplot and code list at the bottom works flawlessly. Many thanks to anyone who can help. ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
[Emc-developers] This run once problem just won't to away.
This gcode, when I can get it to run, finda all 8 holes just fine. And now it will not run twice, refusing to go past the first move command on the 2nd try after a full restart, and rehome.. With no reported errors on the first run.. This is the current copy attached. nuke any others you might have before saving it, I won't touch till someone tells me to. Thanks all. Take care & stay well. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis( linear_bearings.ngc, copyright 25 Feb 2024 by Maurice E Heskett, CET ) #<_measure>=1. ( zero to drill after changing fr probe to 3mm cutter ) #<_hole_spc> = 60. ( nominal space beteen holes drilled ) #<_xy_spd> = 3 #<_z_spd> = 15 #<_search_revs> = 2000 ( first find starting x y z ) G17 G21 G40 G49 G54 G80 G90 G94 ( setup to use xy arcs, metric, ) ( cancel diameter comp,cancel length off set, G54 coordinates system 1, ) ( cancel any canned cycles, absolute distances, feed per minute mode ) ( setup contact to ground to bearing, put probe in spindle with probe wire ) ( NOTE Andy Pugh says this is a valid way to address an array: #[100 + #3] ) ( but he didn't mention that debug was broken, so if we want to use it ) ( to a global var, we must first copy the #<_gvar> = #[basenum+#50] ) ( rather cumbersome proceedure ) (pre-alloocate the bounce vars for floats) o sub (allocate vars) #<_zx> = 0. (for z stuffs) #<_zy> = 0. (for z stuffs) #<_zz> = 0. (for z stuffs) #<_xyx> = 0. (for hole stuff) #<_xyy> = 0. (for hole stuff) #<_xyz> = 0. (for hole stuff) (so first, lets see if it actually works ) #50=0 ( set index to zeroth element ) #[200+#50] = 4 ( zeroth x location for z_check ) #[210+#50] = 1.117 ( zeroth y location for z_check ) #[220+#50] = -66 ( zeroth height to begin z search ) #[230+#50] = 9.5 ( zeroth hole xy x location ) #[240+#50] = 1.117 (zeroth holes xy y location ) #[250+#50] = [-1.75 + #[220+#50]] ( to see if math works) (debug, init_arrays done) o endsub o sub ( now, copy all that to global vars ) #<_zx> = #[200+#50] (z_check x) #<_zy> = #[210+#50] (z_check y) #<_zz> = #[220+#50] (z_check z) #<_xyx> = #[230+#50] (hole_check x) #<_xyy> = #[240+#50] (hole_check y) #<_xyz> = #[250+#50] (hole_check z) o endsub o sub (now the acid test, did it work?) (debug,lb 48 holenum=#50) (debug,lb 49 zx=#<_zx> #<_zy> #<_zz>) (debug,lb 50 xyx=#<_xyx> #<_xyy> #<_xyz>) (debug,lb 51 #100 #101 #102) o endsub o sub ( uses hole arrays 230-240-250 but update all for this hole ) #[230+#50] = #100 (metric x return from hole_finder) #[240+#50] = #101 (metric y return from tholefinder) #[250+#50] = #102 (metric z return from latest g38.2 z-70) o endsub o sub #51 = [#50+1] ( make 51 into next member of array ) #[210+#51] = #101 (propagate current z_c Y) #[220+#51] = #102 (propagate current z_c Z) #[230+#51] = [#100 + #<_hole_spc>] (add hole_spc to x for next hole) #[200+#51] = [#[200+#50] + #<_hole_spc>] (increment z c x spot too) #[240+#51] = #101 ( propagate found y to next hole) #[250+#51] = #102 ( propagate found z to next hole) o endsub o sub (debug,lb zc 72 in z check) s1000 (set spindle rpms) (debug,zc 74 spindle speed set 1000) G1 F1500 z-50 (lift tp clear traveler) G1 f1500 x#<_zx> y#<_xyy> (goto location of z check) G1 F1500 z[#<_zz> +1] (should be a mm above rail) M4 (start spindle, backwards) G4 P.3 (give move time to complete) G38.2 F15 z-70 ( locate top of rail for 1st time) #102 = #5063 ( save real z in mm) (debug,lb zc 82 Z=102=#102) M5 (stop spindle) o endsub o sub (debug,lb hc 87) G1 f1500 z-50 (clear traveler ) G1 f1500 x#<_xyx> y#<_xyy> (position over next hole) G1 f1500 z[#102 -1] (#102 has ben set, should place into hole counterbore) o endsub ( call hole_finder in main loop) o sub (debug,lb hc 94) G1 f1500 z-50 (clear traveler ) G1 f1500 x#100 y#101 (re-position over this hole using found values) G1 f1500 z[#102 -5] (#102 has ben set, should place into small hole) (ready for 2nd call to hole_finder2) o endsub o sub G1 f1500 z-50 G1 f1500 x#<_xyx> y#<_xyy> G1 f1500 z[#<_xyz> +15] (should place above hole) M4 G4 P10 (pause for inspection) M5 o endsub o sub (no z motion now) s#<_search_revs> (debug,HF 116 s/b just into current hole) M4 (start spindle in reverse) G4 p.25 G38.2 F#<_xy_spd> X[#<_xyx>-10] (find left side) G1 F1500 X#<_xyx> Y#<_xyy> (recenter) #110=#5061 (record it) G4 p.25 G38.2 F#<_xy_sp
[Emc-developers] question for Jon Elson
As you know, I have several of you pwm-servo amps. I have never hooked up pin 8-9 because that has not been much of a problem. Probably my fault. I can't see the red led indicating a trip unless I get up a couple steps on a ladder. Room for mice in my garage/shop is sparse so the psu & such are on a top shelf nearly 7 feet up. I was running the spindle at 2 or 3 rpm in high gear yesterday while trying to straighten the wire I use for contact probing, so the pwm was running on an occasional 1 microsecond pulse, no load on the motor, but somehow it tripped and I spent a couple hours, hal printout in hand trying to figure out the why before I remembered that led. Crawled uo to look, sure enough it was on. I had drilled a hole so I could stick a wd-40 stinger in and reset it. So the question is, will a removal of the 12 volts reset it?, and can it be done with the 127 VDC still applied? The 12 volts is never off, currently coming from a 12 volt supply in the interface box that is on anytime the box has power. I read the docs on your site again just now and would interpret the warnings there and interpret that to be a no-no as long as the motor supply has any unused voltage left. That might take 30 minutes or more, that bank of caps is very well formed after a decade within volts of their max rating & there's no other load to bleed it off. I start that motor supply with a 200 watt, 50 ohm resistor in series with the line, shorting it with a 2nd 60 amp SSR 4 seconds later, there is so much capacitance there it trips a 30 amp breaker in the service if not soft-started. I do the same with the line power to the other motor psu's but with a 70 ohm limiter and less delay but that 12 volt in the interface is not on that circuit, so the trip does not get reset by a normal stop/restart. Have you ever given a thought to resetting that with the unused pin 10?. And some hal logic to check the pin 8-9 status, and delay the rest of the power up sequence if it needs reset. The way I am doing this on most of my machines is the machine is cold if F2 has not then toggled on. Many thanks for any advice Jon. Take care & stay well. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Part of my problem with this program
On 3/8/24 13:03, Nicklas SB Karlsson wrote: ... The theory is that even a bent spindle is effectively a perfect circle if it is rotating faster than the servo loop Diameter? Doesn't care as long as the centering move stops the probe noise. If things get really tight around the 12 ga, I run it up against something solid to bend it and increase the push a thou at a time until it runs straight when backed away. pushing too hard fatigues the wire from a continuous micro bend and breaks it of course. This one has been broken a couple times. typically from forgetting the clip lead if the workpiece isn't grounded. Machine frame is grounded to the single point of coarse and this g0704 spindle has cheap steel bearings that ground the spindle. In which case the probe wire hooks to a piece of pcb glueed to some place were I can program a tlo touch off after a tool change. Nicklas Karlsson Take care & stay well Nicklas. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Part of my problem with this program
On 3/8/24 12:46, Nicklas SB Karlsson wrote: tor 2024-03-07 klockan 21:26 -0500 skrev gene heskett: Greetings all; one of the problems I am having when using a reasonable speed to search for the edge of a hole is that it does not stop fast enough on first contact, and leaves the probe touching the side of the hole as it spins often enough I get a probe tripped on the next move which recenters the probe so it can find the next side of the hole. Since the move in all cases is one to open the probe contact, which a .1mm move will do, is there a way to make it ignore such a false trigger during the first few milliseconds of the back to center move? Like disabling the probe input when its not a z move after a g38.2 probe move? Sometimes have or had a similar problem. Guess a good solution would be a distance then moving away from work piece. I have .ngc files to touch off in different directions. Looking into my .ngc files for this there first are G38.n move to probe then probe input is turned off before moving back. Thinking about again this is not a perfect solution, a better option would be to only turn off temporary until part is cleared with some margin but until moved back to start point should not be a problem. To normally have probe turned on should be a good idea to avoid jogging into work piece by misstake, consider this a bug in my .ngc I found then i should write answer for you. Have sometimes seen probes with broken tip for sale. My probe is shop made. A piece of 3/4" teflon rod with a 1/4" tapped hole for a teflon compression nut bore for a 2.5" long piece of stripped wire from a short piece of #12 romex cable as straight as can be manually straightened. Maybe run against something gently to straighten it better so the tip runs pretty true when spinning. Another stranded wire from the probe input is split in 2 and wrapped around the 12 ga to connect the probe input from linuuxcnc And in this case a clip lead from the table to the rail since its all mounted in plastic by a couple of twin screw vise I printed to supply a ground to the rail. violates the heck out of the pick any two out of three rule, simple. expendable and more than accurate enough for the job. And easily repaired in 10 minutes if I should damage it. It doesn't care if its bent half a mm as long as it clears the sidewalls of the hole when reasonably well centered so there is not any contact at the start of a search move. After thinking it thru, verified by halscope the no error when not searching hal thing was being tripped by the blender so I took some timing related stuff back out and slowed the search vel to 10% and it all just works now. That .hal fix was, snipping here and there from the hal file: loadrt and2 names=and_vac,and_lube,and_pid_s_en,steeron,steeroff,and-prb loadrt select8 count=1 addfselect8.0 servo-thread # to control probe sensing addfand-prb servo-thread # fix for broken select8 # # hook up probe signal but disable if motion type not 5 * # # this disables out5 if not a probeing move net probe-ctl <= motion.motion-type => select8.0.sel net probe-sense <= hm2_5i25.0.gpio.030.in_not => and-prb.in0 net probe-hit1 <= select8.0.out5 => and-prb.in1 net probe-hit2 <= and-prb.out => motion.probe-input Simple effective, possibly broken by default blender? IDK. Regards Nicklas Karlsson Take care and stay well, Nicklas Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Part of my problem with this program
On 3/8/24 10:47, andy pugh wrote: On Fri, 8 Mar 2024 at 02:29, gene heskett wrote: is there a way to make it ignore such a false trigger during the first few milliseconds of the back to center move? My probing routines do a fast probe onto the edge (G38.2) and then a slower move out of contact (G38.4) And I keep the result of the G38.4 That would imply a probe switch with some hysteresis. Mine has none. My probe is a copper wire, may not be dead straight, so I run the spindle, usually backwords but because it may not be dead straight, it makes and breaks as the spindle rotates, purposely running the spindle at 2x the servo rate. I re-read the docs just now, and it appears I should be using G38.3, but how does that behave if the move is not achieved, as in the move is too small to find contact? The theory is that even a bent spindle is effectively a perfect circle if it is rotating faster than the servo loop This method is quite accurate, I can get repeat accuracy's 2+ digits right of the decimal point in g21 mode if I search slow enough. Before I do much of anything else I am curious to see how much deviation I have from a dead straight line. Using subroutines makes that easy enough to see. Thanks Andy. Take care & stay well. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Part of my problem with this program
On 3/7/24 21:28, gene heskett wrote: Greetings all; one of the problems I am having when using a reasonable speed to search for the edge of a hole is that it does not stop fast enough on first contact, and leaves the probe touching the side of the hole as it spins often enough I get a probe tripped on the next move which recenters the probe so it can find the next side of the hole. Since the move in all cases is one to open the probe contact, which a .1mm move will do, is there a way to make it ignore such a false trigger during the first few milliseconds of the back to center move? Like disabling the probe input when its not a z move after a g38.2 probe move? Does anybody have a hal solution for what is really a false trigger? Perhaps firing a 10 millisecond disable it timer with the probe trigger? That should give it time to record the #5061-#5063 data and start the clearing move. So I looked at my .hal file, saw what I put in there years ago for that & halscoped what I had for that while running the code you have but with the xy search speed reduced to 2 so I was watching too old paint drying. Worked but halscope showed the 2nd deeper pass was not adequately being recentered leaving a hugely noisy probe signal for pass 2. Also this code ran nearly three times. Since hole_finder calculated new #100 & #101's, I copy pasted hole_finder to hole_finder2 and replaced the guess for x & y center with the previously found #100 and #101. I'm about to go test that version. But I will go see if this version will run twice. And it did, 3 times, sucessfully finding the 3mm hole on the second pass with hole_finder2. I think I'm about ready to change the probe for a 3mm mill and drill some holes. Later today maybe. Thanks for looking at how the old farts brain doesn't work as well as it did 50 years ago. I appreciate it. I still think something in the error code processing is blowing the stack, getting rid of the call within a call seems to have helped, as did slowing the search speed to 10% of the former version. now it never gets more than one call deep. Thanks a bunch everyone, take care & stay well. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
[Emc-developers] Part of my problem with this program
Greetings all; one of the problems I am having when using a reasonable speed to search for the edge of a hole is that it does not stop fast enough on first contact, and leaves the probe touching the side of the hole as it spins often enough I get a probe tripped on the next move which recenters the probe so it can find the next side of the hole. Since the move in all cases is one to open the probe contact, which a .1mm move will do, is there a way to make it ignore such a false trigger during the first few milliseconds of the back to center move? Like disabling the probe input when its not a z move after a g38.2 probe move? Does anybody have a hal solution for what is really a false trigger? Perhaps firing a 10 millisecond disable it timer with the probe trigger? That should give it time to record the #5061-#5063 data and start the clearing move. Ideas welcomed. Thanks all. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Found a new show stopper
On 3/7/24 15:41, Alan Condit via Emc-developers wrote: Gene, In the second block of code following the while loop you’re missing some more “call”s o1 while[#50 le 7] ( stop short ) ( Enf of MAIN WHILE LOOP ) o5 endif o6 if [#<_measure> lt 1] <<<<<<<<<<<<<<< is still 1 o7 do o call o call (see what its got) o call (move to it) (put drill hole call here) #50 = [#50 +1] o7 while [#50 le 7] o6 endif M2 On Mar 7, 2024, at 11:20 AM, emc-developers-requ...@lists.sourceforge.net wrote: o1 while[#50 le 7] ( stop short ) ( Enf of MAIN WHILE LOOP ) o5 endif o6 if [#<_measure> lt 1] o7 do o o (see what its got) o (move to it) (put drill hole call here) #50 = [#50 +1] o7 while [#50 le 7] o6 endif M2 ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Found a new show stopper
On 3/7/24 12:55, John Allwine wrote: Lines 188-190 are missing calls. But they at this time, never execute as #<_measure> is 1, might as well be an M2 above them. On Thu, Mar 7, 2024 at 10:21 AM gene heskett wrote: On 3/7/24 12:12, andy pugh wrote: On Thu, 7 Mar 2024 at 16:44, gene heskett wrote: o It needs to be o CALL What line Andy, call is there in line 160 of the main do/while loop, you've got week old code i think as I've posted two versions. That was from the version you posted 1 hour ago. I can only work with what you provide. See line 163 of this file. I do recall adding the call to that line, a week ago. This one should have a 10 something am date today Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Found a new show stopper
On 3/7/24 12:12, andy pugh wrote: On Thu, 7 Mar 2024 at 16:44, gene heskett wrote: o It needs to be o CALL What line Andy, call is there in line 160 of the main do/while loop, you've got week old code i think as I've posted two versions. That was from the version you posted 1 hour ago. I can only work with what you provide. See line 163 of this file. I do recall adding the call to that line, a week ago. This one should have a 10 something am date today Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ( linear_bearings.ngc, copyright 25 Feb 2024 by Maurice E Heskett, CET ) #<_measure>=1. ( zero to drill after changing fr probe to 3mm cutter ) #<_hole_spc> = 60. ( nominal space beteen holes drilled ) #<_xy_spd> = 15 #<_z_spd> = 25 ( first find starting x y z ) G17 G21 G40 G49 G54 G80 G90 G94 ( setup to use xy arcs, metric, ) ( cancel diameter comp,cancel length off set, G54 coordinates system 1, ) ( cancel any canned cycles, absolute distances, feed per minute mode ) ( setup contact to ground to bearing, put probe in spindle with probe wire ) ( NOTE Andy Pugh says this is a valid way to address an array: #[100 + #3] ) ( but he didn't mention that debug was broken, so if we want to use it ) ( to a global var, we must first copy the #<_gvar> = #[basenum+#50] ) ( rather cumbersome proceedure ) (pre-alloocate the bounce vars for floats) o sub (allocate vars) #<_zx> = 0. (for z stuffs) #<_zy> = 0. (for z stuffs) #<_zz> = 0. (for z stuffs) #<_xyx> = 0. (for hole stuff) #<_xyy> = 0. (for hole stuff) #<_xyz> = 0. (for hole stuff) (so first, lets see if it actually works ) #50=0 ( set index to zeroth element ) #[200+#50] = 4 ( zeroth x location for z_check ) #[210+#50] = 1.117 ( zeroth y location for z_check ) #[220+#50] = -66 ( zeroth height to begin z search ) #[230+#50] = 9.5 ( zeroth hole xy x location ) #[240+#50] = 1.117 (zeroth holes xy y location ) #[250+#50] = [-1.75 + #[220+#50]] ( to see if math works) (debug, init_arrays done) o endsub o sub ( now, copy all that to global vars ) #<_zx> = #[200+#50] (z_check x) #<_zy> = #[210+#50] (z_check y) #<_zz> = #[220+#50] (z_check z) #<_xyx> = #[230+#50] (hole_check x) #<_xyy> = #[240+#50] (hole_check y) #<_xyz> = #[250+#50] (hole_check z) o endsub o sub (now the acid test, did it work?) (debug,lb 48 holenum=#50) (debug,lb 49 zx=#<_zx> #<_zy> #<_zz>) (debug,lb 50 xyx=#<_xyx> #<_xyy> #<_xyz>) (debug,lb 51 #100 #101 #102) o endsub o sub ( uses hole arrays 230-240-250 but update all for this hole ) #[230+#50] = #100 (metric x return from hole_finder) #[240+#50] = #101 (metric y return from tholefinder) #[250+#50] = #102 (metric z return from latest g38.2 z-70) o endsub o sub #51 = [#50+1] ( make 51 into next member of array ) #[210+#51] = #101 (propagate current z_c Y) #[220+#51] = #102 (propagate current z_c Z) #[230+#51] = [#100 + #<_hole_spc>] (add hole_spc to x for next hole) #[200+#51] = [#[200+#50] + #<_hole_spc>] (increment z c x spot too) #[240+#51] = #101 ( propagate found y to next hole) #[250+#51] = #102 ( propagate found z to next hole) o endsub o sub (debug,lb zc 72 in z check) s1000 (set spindle rpms) (debug,zc 74 spindle speed set 1000) G1 F1500 z-50 (lift tp clear traveler) G1 f1500 x#<_zx> y#<_xyy> (goto location of z check) G1 F1500 z[#<_zz> +1] (should be a mm above rail) M4 (start spindle, backwards) G4 P.3 (give move time to complete) G38.2 F15 z-70 ( locate top of rail for 1st time) #102 = #5063 ( save real z in mm) (debug,lb zc 82 Z=102=#102) M5 (stop spindle) o endsub o sub (debug,lb hc 87) G1 f1500 z-50 (clear traveler ) G1 f1500 x#<_xyx> y#<_xyy> (position over next hole) G1 f1500 z[#102 -1] (#102 has ben set, should place into hole counterbore) o endsub ( call hole_finder in main loop) o sub (debug,lb hc 94) G1 f1500 z-50 (clear traveler ) G1 f1500 x#100 y#101 (re-position over this hole using found values) G1 f1500 z[#102 -4] (#102 has ben set, should place into small hole) (ready for 2nd call to hole_finder) o endsub o sub G1 f1500 z-50 G1 f1500 x#<_xyx> y#<_xyy> G1 f1500 z[#<_xyz> +15] (should place above hole) M4 G4 P10 (pause for inspection) M5 o endsub o sub (no z motion now) (debug,HF 113 s/b just into current hole) M4 (start spindle in reverse) G4 p.25 G38.2 F#<_xy_spd> X[#<_xyx>-10] (find left side) G1 F1500 X#<_xyx> Y#<_xyy> (recenter) #110=#5061 (record it) G4 p.2
Re: [Emc-developers] Found a new show stopper
On 3/7/24 11:23, andy pugh wrote: As has been said, you are not calling the subroutines. I am surprised that the code even runs. You have o It needs to be o CALL What line Andy, call is there in line 160 of the main do/while loop, you've got week old code i think as I've posted two versions. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Found a new show stopper
On 3/7/24 09:38, gene heskett wrote: See below> On 3/7/24 05:03, andy pugh wrote: On Thu, 7 Mar 2024 at 00:30, gene heskett wrote: If there is any error that stops it, such as a probe trip when not in a probe move, its all done until shut down, restarted and rehomed. It really shouldn't need a shutdown and restart. It did Andy, then I found a note about subroutines in 11.7.4 that is a bit confusing. First, one named sub can't call another, but then it says it can call itself. So I broke it up into single pieces w/o any nesting in the main loop, but haven't gone out to test it yet as the weather is a bit damp. Probably more typu's to fix when I do. What I'm tryimg to do is locate every 3rd bolthole in a 540mm long 9mm wide linear bearing which is clamped in place by printed jigs on a 560mm long cf tube measureing 20x20, held in good y alignment by two printed vises with opposing movable jaws so it dialed within a thou of straight with the mills table. First I go into the counterbore of the screw hole, locate that with a spinning probe, then if that works center it and go in 4 or 5mm to get inside the 3mm hole, find that and record xyz in the array. Find the center of the 3mm hole and re-record that. then increment the x in both arrays by 60mm and find that hole, wash rinse repeat for 8 holes. If I can do that, then I change the probe for a 3mm sc mill and drill the holes in the cf for 3mm bolts. Then turn it over in the vises and do it again since I put another rail on the bottom of the cf tube. The printer I'm rebuilding, a tronxy400pro that never worked had all that flying weight made out of steel and solid alu and is going to lose at least half a pound of flying weight the xy motors won't have to throw around...The plain steppers have also been replaced with hanpose stepper/servos running on 3x the normal driver voltage, with optical encoders feeding the matching drivers. Turning a 40mm/second printer into a 400mm/second+ printer. I hope. To better describe it, if the code is edited and reloaded after an error, all the stuff before the first move that has debugging statements in it, executes normally but it will not execute past the first machine move on the subsequent try until stopped and restarted, rehomed, and the code reloaded. Weirdsville... Even w/o errors, it only works once. Thanks Andy. Take care & stay well. Cheers, Gene Heskett, CET. I must be doing something wrong. Despite the rewrite, if there is an error, it runs just once and will not to run again, stopping in o at the 1st machine move, then must be shut down, restarted and rehomed, and reloaded. Backplot looks great, no error reported on 2nd reload and run, but won't move machine. gcode attached. Something I'm doing wrong? I can rehome it, move the machine from the keyboard, but the gcode can't move it on the 2nd run. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ( linear_bearings.ngc, copyright 25 Feb 2024 by Maurice E Heskett, CET ) #<_measure>=1. ( zero to drill after changing fr probe to 3mm cutter ) #<_hole_spc> = 60. ( nominal space beteen holes drilled ) #<_xy_spd> = 15 #<_z_spd> = 25 ( first find starting x y z ) G17 G21 G40 G49 G54 G80 G90 G94 ( setup to use xy arcs, metric, ) ( cancel diameter comp,cancel length off set, G54 coordinates system 1, ) ( cancel any canned cycles, absolute distances, feed per minute mode ) ( setup contact to ground to bearing, put probe in spindle with probe wire ) ( NOTE Andy Pugh says this is a valid way to address an array: #[100 + #3] ) ( but he didn't mention that debug was broken, so if we want to use it ) ( to a global var, we must first copy the #<_gvar> = #[basenum+#50] ) ( rather cumbersome proceedure ) (pre-alloocate the bounce vars for floats) o sub (allocate vars) #<_zx> = 0. (for z stuffs) #<_zy> = 0. (for z stuffs) #<_zz> = 0. (for z stuffs) #<_xyx> = 0. (for hole stuff) #<_xyy> = 0. (for hole stuff) #<_xyz> = 0. (for hole stuff) (so first, lets see if it actually works ) #50=0 ( set index to zeroth element ) #[200+#50] = 4 ( zeroth x location for z_check ) #[210+#50] = 1.117 ( zeroth y location for z_check ) #[220+#50] = -66 ( zeroth height to begin z search ) #[230+#50] = 9.5 ( zeroth hole xy x location ) #[240+#50] = 1.117 (zeroth holes xy y location ) #[250+#50] = [-1.75 + #[220+#50]] ( to see if math works) (debug, init_arrays done) o endsub o sub ( now, copy all that to global vars ) #<_zx> = #[200+#50] (z_check x) #<_zy> = #[210+#50] (z_check y) #<_zz> = #[220+#50] (z_check z) #<_xyx> = #[230+#50]
Re: [Emc-developers] Found a new show stopper
On 3/7/24 05:03, andy pugh wrote: On Thu, 7 Mar 2024 at 00:30, gene heskett wrote: If there is any error that stops it, such as a probe trip when not in a probe move, its all done until shut down, restarted and rehomed. It really shouldn't need a shutdown and restart. It did Andy, then I found a note about subroutines in 11.7.4 that is a bit confusing. First, one named sub can't call another, but then it says it can call itself. So I broke it up into single pieces w/o any nesting in the main loop, but haven't gone out to test it yet as the weather is a bit damp. Probably more typu's to fix when I do. What I'm tryimg to do is locate every 3rd bolthole in a 540mm long 9mm wide linear bearing which is clamped in place by printed jigs on a 560mm long cf tube measureing 20x20, held in good y alignment by two printed vises with opposing movable jaws so it dialed within a thou of straight with the mills table. First I go into the counterbore of the screw hole, locate that with a spinning probe, then if that works center it and go in 4 or 5mm to get inside the 3mm hole, find that and record xyz in the array. Find the center of the 3mm hole and re-record that. then increment the x in both arrays by 60mm and find that hole, wash rinse repeat for 8 holes. If I can do that, then I change the probe for a 3mm sc mill and drill the holes in the cf for 3mm bolts. Then turn it over in the vises and do it again since I put another rail on the bottom of the cf tube. The printer I'm rebuilding, a tronxy400pro that never worked had all that flying weight made out of steel and solid alu and is going to lose at least half a pound of flying weight the xy motors won't have to throw around...The plain steppers have also been replaced with hanpose stepper/servos running on 3x the normal driver voltage, with optical encoders feeding the matching drivers. Turning a 40mm/second printer into a 400mm/second+ printer. I hope. To better describe it, if the code is edited and reloaded after an error, all the stuff before the first move that has debugging statements in it, executes normally but it will not execute past the first machine move on the subsequent try until stopped and restarted, rehomed, and the code reloaded. Weirdsville... Even w/o errors, it only works once. Thanks Andy. Take care & stay well. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Found a new show stopper
On 3/6/24 15:49, gene heskett wrote: specifically the command to set spindle speed MUST BE LOWERCASE Nothing in a gcode file causing machine motion beyond S1000 will execute. No error messages of any kind either. Found it by surround the first move with (debug, msg) statements. I have not gone to the new number the spindle syntax because that reference in the docs is not very clear and doesn't actually explain it that well, at least to me. Does it require a $ sign in front of the spindle number? My limited exploration has not been successful at rotating a spindle. Since I don't have anything that does qualify as a spindle although I've both A and B axises now that can turn 400 rpms synched to x or y, I would not call then spindles as their chucks are in fact 3d printed square sockets. The backplot showed the motion trace just fine for either s case. I have killed it a couple times, but changing the view fixes it. I changed the file back to a lowercase s: s1000 and everything then seems to run normally. So the old canary will live. ;o)> But he is certainly surprised. Looks like its more than just what looks like a case sensitivity. What its doing is actually only running the attached gcode once. If there is any error that stops it, such as a probe trip when not in a probe move, its all done until shut down, restarted and rehomed. There are two places where one sub calls another. If there are no errors, it all works. Is that a no-no in terms of error recovery, as in blowing away the call stack? Take care & stay well everyubody. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ( linear_bearings.ngc, copyright 25 Feb 2024 by Maurice E Heskett, CET ) #<_measure>=1. ( zero to drill after changing fr probe to 3mm cutter ) #<_hole_spc> = 60. ( nominal space beteen holes drilled ) #<_xy_spd> = 15 #<_z_spd> = 25 ( first find starting x y z ) G17 G21 G40 G49 G54 G80 G90 G94 ( setup to use xy arcs, metric, ) ( cancel diameter comp,cancel length off set, G54 coordinates system 1, ) ( cancel any canned cycles, absolute distances, feed per minute mode ) ( setup contact to ground to bearing, put probe in spindle with probe wire ) ( NOTE Andy Pugh says this is a valid way to address an array: #[100 + #3] ) ( but he didn't mention that debug was broken, so if we want to use it ) ( to a global var, we must first copy the #<_gvar> = #[basenum+#50] ) ( rather cumbersome proceedure ) (pre-alloocate the bounce vars for floats) o sub (allocate vars) #<_zx> = 0. (for z stuffs) #<_zy> = 0. (for z stuffs) #<_zz> = 0. (for z stuffs) #<_xyx> = 0. (for hole stuff) #<_xyy> = 0. (for hole stuff) #<_xyz> = 0. (for hole stuff) (so first, lets see if it actually works ) #50=0 ( set index to zeroth element ) #[200+#50] = 4 ( zeroth x location for z_check ) #[210+#50] = 1.117 ( zeroth y location for z_check ) #[220+#50] = -66 ( zeroth height to begin z search ) #[230+#50] = 9.5 ( zeroth hole xy x location ) #[240+#50] = 1.117 (zeroth holes xy y location ) #[250+#50] = [-1.75 + #[220+#50]] ( to see if math works) (debug, init_arrays done) o endsub o sub ( now, copy all that to global vars ) #<_zx> = #[200+#50] (z_check x) #<_zy> = #[210+#50] (z_check y) #<_zz> = #[220+#50] (z_check z) #<_xyx> = #[230+#50] (hole_check x) #<_xyy> = #[240+#50] (hole_check y) #<_xyz> = #[250+#50] (hole_check z) o endsub o sub (now the acid test, did it work?) (debug,lb 48 holenum=#50) (debug,lb 49 zx=#<_zx> #<_zy> #<_zz>) (debug,lb 50 xyx=#<_xyx> #<_xyy> #<_xyz>) (debug,lb 51 #100 #101 #102) (debug,show args done) o endsub o sub ( uses hole arrays 230-240-250 but update all for this hole ) (#[200+#50] = #100) (#[210+#50] = #101) (#[220+#50] = #102) #[230+#50] = #100 (metric x return from hole_finder) #[240+#50] = #101 (metric y return from tholefinder) #[250+#50] = #102 (metric z return from latest g38.2 z-70) o endsub o sub #51 = [#50+1] ( make 51 into next member of array ) #[210+#51] = #101 (propagate current Y) #[220+#51] = #102 (propagate current Z) #[230+#51] = [#100 + #<_hole_spc>] (add hole_spc to x for next hole) #[200+#51] = [#[200+#50] + #<_hole_spc>] (increment z check x spot too) #[240+#51] = #101 ( propagate found y) #[250+#51] = #102 ( propagate found z) o endsub o sub (debug,lb zc 76 in z check) s1000 (set spindle rpms) (debug,zc spindle speed set 1000) G1 F1500 z-50 (lift tp clear traveler) (debug, lb zc 80 lifted to -50) G1 f1500 x#<_
[Emc-developers] Found a new show stopper
specifically the command to set spindle speed MUST BE LOWERCASE Nothing in a gcode file causing machine motion beyond S1000 will execute. No error messages of any kind either. Found it by surround the first move with (debug, msg) statements. I have not gone to the new number the spindle syntax because that reference in the docs is not very clear and doesn't actually explain it that well, at least to me. Does it require a $ sign in front of the spindle number? My limited exploration has not been successful at rotating a spindle. Since I don't have anything that does qualify as a spindle although I've both A and B axises now that can turn 400 rpms synched to x or y, I would not call then spindles as their chucks are in fact 3d printed square sockets. The backplot showed the motion trace just fine for either s case. I have killed it a couple times, but changing the view fixes it. I changed the file back to a lowercase s: s1000 and everything then seems to run normally. So the old canary will live. ;o)> But he is certainly surprised. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
[Emc-developers] Need a howto
I have used the attached file yonks ago, worked fine, used to determine
which mode, inch or mm the machine is in. But now halcompile will not
build it. I've tried all the options to halcompile, but nothing works.
It seems all roads lead to the same exit:
gene@GO704:~/linuxcnc/configs/go704-xyza$ halcompile --compile units.comp
Traceback (most recent call last):
File "/usr/bin/halcompile", line 1558, in
main()
File "/usr/bin/halcompile", line 1528, in main
process(f, mode, outfile)
File "/usr/bin/halcompile", line 1366, in process
a, b = parse(filename)
File "/usr/bin/halcompile", line 441, in parse
a, b = f.split("\n;;\n", 1)
ValueError: not enough values to unpack (expected 2, got 1)
It is older code from 2017, by Kurt Jacobsen
Can anyone advise?
Cheers, Gene Heskett, CET.
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author, 1940)
If we desire respect for the law, we must first make the law respectable.
- Louis D. Brandeis#!/usr/bin/python
# Author: Kurt Jacobson
#Date: 04/15/17
# Email: kurtjacob...@bellsouth.net
# License: GPL
import hal, time
import linuxcnc
stat = linuxcnc.stat()
h = hal.component("units")
h.newpin("prog_in", hal.HAL_BIT, hal.HAL_OUT)
h.newpin("prog_mm", hal.HAL_BIT, hal.HAL_OUT)
h.newpin("update_rate", hal.HAL_FLOAT, hal.HAL_IN)
h.ready()
# Set default update rate to 50ms
h.update_rate = .05
try:
while 1:
time.sleep(h.update_rate)
stat.poll()
prog_units = stat.program_units
if prog_units == 1: # Inch
h.prog_in = 1
h.prog_mm = 0
elif prog_units == 2: # Metric
h.prog_in = 0
h.prog_mm = 1
except KeyboardInterrupt:
raise SystemExit
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] I'll bite. array's don't work write only
On 3/3/24 20:01, John Allwine wrote: If you’re testing with MDI commands, either a local or global variable would work. If you have a subroutine where you’re setting the value and then writing an MDI command to print the value, then it would have to be a global variable because a local variable would go out of scope. That is ok, for this I'd rather use globals with unique names. It never occured to me the debug might be broken and lieing to me. What is the code that isn’t reliably working? In your example, it’s only the (debug,) statements that aren’t behaving how you expect. The debug command knows how to replace variable names, but doesn’t know how to evaluate an expression. Assign the expression to a variable and then print that variable and it will do what you want. If your variable goes out of scope, though, before printing it then it won’t show anything in the variable. I'm aware of that, John, that is the reason I asked the question. I've not started on the re-write yet. I have 2 sets of arrays, one for finding the top z beside the predrilled hole, and another set of three to place the probe somewhere near the center of the hole and 2mm deeper than the z for finding the xy. Electrical contact, spindle running so the probe is dead circular. I have even done it with a tool spinning backwards, does not mark a pcb. Thank you, John. Take care & Stay well. On Mar 3, 2024, at 4:56 PM, gene heskett wrote: On 3/3/24 17:33, John Allwine wrote: Store it in another variable first: # = #[200+#50] ;debug,# That is what I am trying to do, prolly 20+ times in my present code but regardless # contains all balls. and its global here, 3 other machines including the rpi4b running my Sheldon are also behaving identically. Works fine on the left of the = sign does something else when its on the right side of the = sign. I did get it to work but probably after a dozen tries. Dependably? TBD. And I'm frazzled for this day. So I'll give it a try tom. Thanks John, take care & stay well. ?? Does the local/global status of the var make a diff? I note you used a local in your example above. Thanks a bunch. [...] Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] I'll bite. array's don't work write only
On 3/3/24 17:33, John Allwine wrote: Store it in another variable first: # = #[200+#50] ;debug,# That is what I am trying to do, prolly 20+ times in my present code but regardless # contains all balls. and its global here, 3 other machines including the rpi4b running my Sheldon are also behaving identically. Works fine on the left of the = sign does something else when its on the right side of the = sign. I did get it to work but probably after a dozen tries. Dependably? TBD. And I'm frazzled for this day. So I'll give it a try tom. Thanks John, take care & stay well. ?? Does the local/global status of the var make a diff? I note you used a local in your example above. Thanks a bunch. [...] Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
[Emc-developers] I'll bite. array's don't work write only
preset #50=0 #[200+#50]= random decimal value, like 777.7 the above syntax does not report any error in mdi line (debug, #200) returns 777.7 (debug, #[200+#50]) gets me "#[200+0.00]" So how do I read it back? A bare #200 works but #[200+0] doesn't. #[200+1]=. (debug, #201) returns . (debug , #[200+1]) returns "#[200+1]" but not the value stored there. So how do I read back what I wrote? Or is it busted? Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Dynamically building a string for a varname?
On 3/2/24 10:28, Andy Pugh wrote:
On 2 Mar 2024, at 12:35, gene heskett wrote:
I renamed tholefinder.ngc to tholefinder,
Leave the filename the same.
Rename the subroutine inside that file.
Call it by the subroutine name.
I did that, it worked but now complains about a syntax error in that
file. But without reporting a line number or anything that might lead to
locating it. And I have another problem, tholefinder was written in inch
days, last used for inch measurements. To make it smarter I need to have
it check the machines units with units.comp, which I finally found a
copy of but halcompile flat refuses to do anything with it. Every combo
the manpage shows exits with:
gene@GO704:~/linuxcnc/configs/go704-xyza$ halcompile --install units.comp
Traceback (most recent call last):
File "/usr/bin/halcompile", line 1558, in
main()
File "/usr/bin/halcompile", line 1528, in main
process(f, mode, outfile)
File "/usr/bin/halcompile", line 1366, in process
a, b = parse(filename)
File "/usr/bin/halcompile", line 441, in parse
a, b = f.split("\n;;\n", 1)
ValueError: not enough values to unpack (expected 2, got 1)
units.comp file attached
Is the man page for halcompile duff? This is master, apt updated early
this morning. On buster of course.
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers
.
Thank you.
Cheers, Gene Heskett, CET.
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author, 1940)
If we desire respect for the law, we must first make the law respectable.
- Louis D. Brandeis
#!/usr/bin/python
# Author: Kurt Jacobson
#Date: 04/15/17
# Email: kurtjacob...@bellsouth.net
# License: GPL
import hal, time
import linuxcnc
stat = linuxcnc.stat()
h = hal.component("units")
h.newpin("prog_in", hal.HAL_BIT, hal.HAL_OUT)
h.newpin("prog_mm", hal.HAL_BIT, hal.HAL_OUT)
h.newpin("update_rate", hal.HAL_FLOAT, hal.HAL_IN)
h.ready()
# Set default update rate to 50ms
h.update_rate = .05
try:
while 1:
time.sleep(h.update_rate)
stat.poll()
prog_units = stat.program_units
if prog_units == 1: # Inch
h.prog_in = 1
h.prog_mm = 0
elif prog_units == 2: # Metric
h.prog_in = 0
h.prog_mm = 1
except KeyboardInterrupt:
raise SystemExit
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Dynamically building a string for a varname?
On 3/2/24 04:59, andy pugh wrote: On Fri, 1 Mar 2024 at 23:38, gene heskett wrote: Well, I seem to be up to the first named subroutine call. But it cannot find it. Drop the ".ngc" from the subroutine name in the .ngc fle. You have o sub Didn't help, Andy. I renamed tholefinder.ngc to tholefinder, fixed it internally, It is not (most of the time) looking in the right place as specified in [RS@&$NGC] SUBROUTINE_PATH in the .ini file. See attached screenshot. Most of the time its looking in /tmp, /home/tmp (which doesn't exist, or /home/gene/tmp which is apparently where its caching stuff. but NEVER looks in /home/gene/linuxcnc/nc_files/subs/ which is where [RS274NGC} SUBROUTINE_PATH = ~/linuxcnc/nc_files/subs point it at. Am I being dense? Or are you saying the files name still has the .ngc, and the search adds that to the simplified name used in the parent file? I'll try that but after some shoe-goo cures a while. After a decade the glue holding the gear shift tally switches has given way so I had to reglue that assembly before the spindle runs right. Takes about an hour. Thank you, take care & stay well. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Dynamically building a string for a varname?
On 3/1/24 07:22, andy pugh wrote: On Fri, 1 Mar 2024 at 08:04, gene heskett wrote: One last Q in this thread, assuming the machine has been homed and I want to G38.2 z-70. where contact is expected at about -67mm. But I occasionally get a stop before contact. Noise? Will #5070 tell me?. No, the system can't tell the difference between a false and a genuine trigger. Well, I seem to be up to the first named subroutine call. But it cannot find it. my .ini file has: [RS274NGC] DISABLE_G92_PERSISTENSE = 0 PARAMETER_FILE = linuxcnc.var SUBROUTINE_PATH = /home/gene/linuxcnc/nc_files/subs/ RS274NGC_STARTUP_CODE = G21 G61 #4990=0. #4991=0. #4992=0. #4993=0. #4994=0. #4995=0.00 Then I noted that it wasn't looking in the directory named above, but in a hash named subdir in /tmp. So I hunted it down with mc and deleted the whole thing from /tmp, I had restarted and rehomed lcnc a couple times to get the stripped to the above path to subs into effect but after nukeing it from /tmp the EOF error is gone. Does anyone have any idea where/how/why /tmp even got into the lcnc env? 4 damned days I've looked at that error, seeing the filename and the line # of the error w/o a solution. Stopped lcnc, nuked that dir in /tmp restarted linuxcnc, the dir in /tmp is recreated but it still cannot find that file, and the path to it is not being looked at despite the above SUBROUTINE_PATH being specified, so I moved the file to nc_files where the error msg says it is looking, but still can't find it, put copies both places, still can't find it, copy of tholefinder.ngc file attached. Whats wrong? I'm DIW. Thanks all. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis o sub ( tholefinder.ngc, a hole finding routine to aid in properlyregistering ) ( the top and bottom machining operations based on the eagle CAD program. ) ( The files are issued by pcb2gcode ulp when it converts an eagle designed ) ( board to gcode commands) ( VERY IMPORTANT this code makes some assumptions ) ( I am also rigging it to look at a much larger hole, in this case about ) ( a 27mm hole in the bellows nipples for the Sheldon's z screw bellows covers ) ( 1. That a small piece of brass tubing has been let into the pallet, at ) ( board left front corner -0.2X and +0.1Y and is connected to the probe ) ( circuit ) ( a year later, working on Kip Koons first board ) ( now effectively random, and we must find that offset anew ) ( when this is finalized, copy to coco-baord1 directory ) ( 2. This location device is not known to eagle. Since I found that the ) ( whole pattern worked better if the eagle output files were moved to ) ( the left by 0.1X, this program will leave at $100 in memory, a value ) ( that when applied with a G43.1 X#100 command, the offset to X that will ) ( do that, but _all_ tool probing is done without that offset in effect ) ( 3. The mill is driven by gcode or by hand before this routine is run ) ( so that the tip of a conical probe is already reasonably well centered ) ( and lowered slightly into the hole and that the probes hot lead is ) ( actually connected to the board! ) ( 4. From that point we can probe to find the exact X center of the hole ) ( in just 3 probing moves, and an additional 2 probing moves will show ) ( the actual Y offset. For informational use only so far. ) ( 5. My probe for this was a very small dremel burr that was cut off to ) ( leave a tapered shank but it has some runout, hence the probe is turning ) ( slowly at high rate in comparison to the probing moves to effectively ) ( cancel out the runout, improving the accuracy by at least 2 decimal places) ( 6. The returned value will be returned in var #100, using other 100 up ) ( vars for local scratchpad as it runs & available for bot.etch and ) ( bot.drill use in setting a G43.1 offset ) ( 7. In any event this MUST be used before any other top machining ) ( operations. ) ( Copyright Feb 2012 by Gene Heskett, licensed the same as LinuxCNC ) ( these values are custom for each board pallet made ) ( change signs of fudges here in these 2 statements ) ( set initial search limits ) (and I just found why some puzzliing results, I was re-using inital values! ) ( so use tmp_name for contact results ) #<_right_x> = 0.150 #<_1st_speed> = 1.5 #<_2nd_speed> = .05 #<_left_x> = -#<_right_x> #<_back_y> = #<_right_x> #<_front_y> = -#<_right_x> #<_tmp_f_y> = 0.000 #<_tmp_b_y> = 0.000 #<_tmp_l_x> = 0.000 #<_tmp_r_x> = 0.000 ( Assume machine is lost, reset so machine & g54 match ) ( G53
Re: [Emc-developers] Dynamically building a string for a varname?
On 2/29/24 18:50, gene heskett wrote: On 2/29/24 15:56, andy pugh wrote: On Thu, 29 Feb 2024 at 18:14, gene heskett wrote: Is it possible to concatenate several string vars into a variable name in order to access an array of vars No. But you can do this with numbered variables. So #[100 + #3] is a valid construction. tholefinder.ngc currently uses #100 and #101 to return the exact xy location of the bolthole its positioned to check (but that is adjustable) so I likely will use #200, #210, #220 for the anchor points of the 3 0-9 element arrays for x,y,z. Any possible interference with o### etc while/do/etc markers? How do I initialize that to 0., the same way, I assume by #300=0, then #3000=first hole location x while incremnting #300 to 9 #3010=y location etc, #3020=zlocattion. Just to make sure its reserved when I write the actual data returned from tholefinder (which returns the exact center of the holes x in #100 y=in #101 etc) to that location for later use. This will greatly simplify the code, thank you very much Andy. Even simpler if I don't have to init the storage until I actually have data to write to that location. One last Q in this thread, assuming the machine has been homed and I want to G38.2 z-70. where contact is expected at about -67mm. But I occasionally get a stop before contact. Noise? Will #5070 tell me?. The docs for G38.2 should say if the value z is searching for is absolute or relative from whence it starts. Needs a 1 liner to add that missing data. Docs are 2.9.0-pre1+git20230208.f1270d6ed7 so pretty old now. but bookworm isn't offering me anything newer? shouldn't it be up to 2.9.2 now? Thanks Andy. Stay warm & well. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Dynamically building a string for a varname?
On 2/29/24 15:56, andy pugh wrote: On Thu, 29 Feb 2024 at 18:14, gene heskett wrote: Is it possible to concatenate several string vars into a variable name in order to access an array of vars No. But you can do this with numbered variables. So #[100 + #3] is a valid construction. tholefinder.ngc currently uses #100 and #101 to return the exact xy location of the bolthole its positioned to check (but that is adjustable) so I likely will use #200, #210, #220 for the anchor points of the 3 0-9 element arrays for x,y,z. Any possible interference with o### etc while/do/etc markers? How do I initialize that to 0., the same way, I assume by #300=0, then #3000=first hole location x while incremnting #300 to 9 #3010=y location etc, #3020=zlocattion. Just to make sure its reserved when I write the actual data returned from tholefinder (which returns the exact center of the holes x in #100 y=in #101 etc) to that location for later use. This will greatly simplify the code, thank you very much Andy. Even simpler if I don't have to init the storage until I actually have data to write to that location. Thanks Andy. Stay warm & well. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
[Emc-developers] side question
On these chinese linear bearing rails, are the boltholes drilled in two ops. the shcs hear socket, then the bolt hole on thru, or with a custom cutter all in one step? My probe is a 12 gauge copper wire, spinning at 300 revs in the spindle and might have enough tip wobble to prevent it from descending into the 3mm bolt hole to actually measure the bolthole so I may have to settle for centering in the SHCS counterbore. And assume the bolthole is centered in the SHCS clearance. My experience has been they're a lot sloppier at the bolt hole locations than in bearing precision which is pretty good. What is your experience in similar situations? Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
[Emc-developers] Dynamically building a string for a varname?
Good afternoon; Is it possible to concatenate several string vars into a variable name in order to access an array of vars by appending one or more chars to build the name of an array of values? Something like starting with the general array name like the first 5 chars of "#<_ar" then append the axis letter, then append the while loops control variable 0 to 9 and add a final ">" to terminate the name of the var winding up with #<_arx4>, in this case meaning the 5th x location storage name, assuming a base0 numbering system. I've spent several hours scanning thru the docs pdf w/o finding any string manipulation stuff. It would be handier than sliced bread or bottled beer for this. And would shrink my final code to 15% of unrolling the whole maryann bolthole by bolthole. Thanks all. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] got usability problem w/master
On 2/29/24 04:05, andy pugh wrote: On Wed, 28 Feb 2024 at 17:14, gene heskett wrote: The Halshow "Watch" tab can be better for this. Sounds useful. is it covered by a manpage? https://linuxcnc.org/docs/stable/html/hal/tools.html#_halshow Bookmarked, thanks Andy. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] got usability problem w/master
On 2/29/24 04:03, andy pugh wrote: On Wed, 28 Feb 2024 at 21:29, gene heskett wrote: and I've copies of it in nc_files and in nc_files/subs, error msg claims EOF while searching for it. That isn't the most helpful message, it means that it hasn't found the subroutine in the current G-code file (and had got to the end of the file looking) What this normally means is that the subroutine file has not been found in the subroutine path. Working out why can be difficult. I think I have that sorted Andy, deep in the docs is a mention that it need the "o sub" at the head of the file, an ends with "o endsub", and I had never added that stuff to the file. So I did now. So that s/b fixed. But I was getting wet running back and forth to the machine as it rained fairly steadily all day yesterday Since theres a bucket of holes to be drilled, I have setup a slew of vars, #<_arx, #<-ary, #<_arz with a single digit hex value as the next character and of coarse a closing ">. I'll only drill every 2nd or 3rd hole as this is the X axis for a 3d printer, 490mm long. 9mm wide linear bearings top and bottom of a 540mm long 20x20 cf tube. Puting flying weight on a starvation diet, printing parts that were originally poorly designed and made from 8mm steel plate! Now, I know bash can do it, I've done it decades ago in amigados and either bash or ARexx, but can lcnc concatenate 4 strings, starting with "#<_ar" one the axises(x,y,z), as the 2rd arg, the while counter value in hex and the 4th the ">" in order to dynamically generate the var name "#<_arx5>" that can then be used to address that particular variable? That would allow me to write this as a while loop as opposed to unrolling this into 1600 LOC. Most of which is copy, paste, edit. If lcnc can do this, what section discusses the howto? Thanks Andy. Take care & stay well. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] got usability problem w/master
On 2/28/24 16:27, gene heskett wrote: On 2/28/24 12:15, gene heskett wrote: On 2/28/24 08:02, andy pugh wrote: On Wed, 28 Feb 2024 at 11:59, gene heskett wrote: I often have as many as 8 open while developing code so I can see whats going on in real time and now I can't see things w/o clicking on them, The Halshow "Watch" tab can be better for this. You can build a list, (and save the layout) and can also set the values of inputs, as well as view outputs. Sounds useful. is it covered by a manpage? What I am attempting to do would be useful to have an array of xy locations, but how does one go about defining an #<_arraynamex+index> and #<_arraynamey+index>? The first portion of this particular program is to survey the locations of the hole the holes I'm actually going to put bolts in, and I figure since this is not going to be overly flexible, is to locate the 8 holes (every third one) and if I cannot use the array by indexing into it my code will grow like Topsy cuz I'll have to write it 8 times instead of looping thru it once. Reading thru the docs it appears the "g1 f2 x[#<_arx>+[#100*#<_distance>]] y#<_ary>" might be legal. That would let me write a knowing only the distance from hole 0 to hole 7 by finding those two holes and dividing by 7. And would only need 3 storage locations for x (assuming the bolt holes were accurately bored. Rarely the case with chinese linear bearings. I swear their boltholes are drilled by hand. I have one that y wanders so far off center that I can see the countersink in the bottom of the ball track. So unless you can tell me how to set up an 8 element xy array, I'll have to unroll the loop and use at least 18 vars. Can you rescue me? Thanks Andy. Now I have a different problem. A subroutine I wrote back in 2012 has ceased to load correctly using this syntax. o call Ack the docs it must be named internally as a sub and again as an endsub to match the call in the parent program in addition to being named the same as the call? Took me hours to find that in the docs. Whatever, I guess it shows how long its been since I did use it last. But it will work hole by hole. Long day, back at it tom. Thanks all. and I've copies of it in nc_files and in nc_files/subs, error msg claims EOF while searching for it. Yet I've used it hundreds of times to find a hole so I can rigid tap it. Thanks for any hints. Cheers, Gene Heskett, CET. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] got usability problem w/master
On 2/28/24 12:15, gene heskett wrote: On 2/28/24 08:02, andy pugh wrote: On Wed, 28 Feb 2024 at 11:59, gene heskett wrote: I often have as many as 8 open while developing code so I can see whats going on in real time and now I can't see things w/o clicking on them, The Halshow "Watch" tab can be better for this. You can build a list, (and save the layout) and can also set the values of inputs, as well as view outputs. Sounds useful. is it covered by a manpage? What I am attempting to do would be useful to have an array of xy locations, but how does one go about defining an #<_arraynamex+index> and #<_arraynamey+index>? The first portion of this particular program is to survey the locations of the hole the holes I'm actually going to put bolts in, and I figure since this is not going to be overly flexible, is to locate the 8 holes (every third one) and if I cannot use the array by indexing into it my code will grow like Topsy cuz I'll have to write it 8 times instead of looping thru it once. Reading thru the docs it appears the "g1 f2 x[#<_arx>+[#100*#<_distance>]] y#<_ary>" might be legal. That would let me write a incrementing #100 from 0 to 7 knowing only the distance from hole 0 to hole 7 by finding those two holes and dividing by 7. And would only need 3 storage locations for x (assuming the bolt holes were accurately bored. Rarely the case with chinese linear bearings. I swear their boltholes are drilled by hand. I have one that y wanders so far off center that I can see the countersink in the bottom of the ball track. So unless you can tell me how to set up an 8 element xy array, I'll have to unroll the loop and use at least 18 vars. Can you rescue me? Thanks Andy. Now I have a different problem. A subroutine I wrote back in 2012 has ceased to load correctly using this syntax. o call and I've copies of it in nc_files and in nc_files/subs, error msg claims EOF while searching for it. Yet I've used it hundreds of times to find a hole so I can rigid tap it. Thanks for any hints. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] got usability problem w/master
On 2/28/24 09:35, John Thornton wrote: I'm using Debian with the Mate desktop and I use the TraditionalOk theme. It's not perfect but better than the rest. I just went out and changed the apperannce theme several times with no noticable effect. buster yet, xfce4 desktop. Thanks John. JT On 2/28/2024 5:59 AM, gene heskett wrote: On 2/28/24 06:27, John Thornton wrote: That's probably the OS theme doing that. I ran into that with Debian a few years ago the theme updated then windows not in focus became very dim. It was so annoying I searched for a new theme that didn't do that. Do you recall which one fixed this? Thanks John. Stay well. JT On 2/27/24 19:12, gene heskett wrote: Hey guys; Genuine show stopper here at the Heskett den; Running an axis gui. linuxcnc master updated yesterday. Camera gone again. Has been gone for months. Minor, this project won't use it anyway. I've almost given up on ever having a dependable camera. Some one since the last time I wrote some gcode on my g0704, has hooked the hal meter up to focus status, so if it not focused, its so dim I can't read it unless I'm under a foot from its 21" monitor, so when I click on axis so the keyboard can move he machine to manually verify the probe is working, I can't read it, unfocused, its way too dim. halmeter and halscope if its affected, so they s/b be removed from focus and made to behave as if they are focused full time. I am DIW until i can see this stuff. Is there anything I can do in hal to fix this? Thanks all. Cheers, Gene Heskett, CET. ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett, CET. ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] got usability problem w/master
On 2/28/24 08:02, andy pugh wrote: On Wed, 28 Feb 2024 at 11:59, gene heskett wrote: I often have as many as 8 open while developing code so I can see whats going on in real time and now I can't see things w/o clicking on them, The Halshow "Watch" tab can be better for this. You can build a list, (and save the layout) and can also set the values of inputs, as well as view outputs. Sounds useful. is it covered by a manpage? What I am attempting to do would be useful to have an array of xy locations, but how does one go about defining an #<_arraynamex+index> and #<_arraynamey+index>? The first portion of this particular program is to survey the locations of the hole the holes I'm actually going to put bolts in, and I figure since this is not going to be overly flexible, is to locate the 8 holes (every third one) and if I cannot use the array by indexing into it my code will grow like Topsy cuz I'll have to write it 8 times instead of looping thru it once. Reading thru the docs it appears the "g1 f2 x[#<_arx>+[#100*#<_distance>]] y#<_ary>" might be legal. That would let me write a incrementing #100 from 0 to 7 knowing only the distance from hole 0 to hole 7 by finding those two holes and dividing by 7. And would only need 3 storage locations for x (assuming the bolt holes were accurately bored. Rarely the case with chinese linear bearings. I swear their boltholes are drilled by hand. I have one that y wanders so far off center that I can see the countersink in the bottom of the ball track. So unless you can tell me how to set up an 8 element xy array, I'll have to unroll the loop and use at least 18 vars. Can you rescue me? Thanks Andy. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] got usability problem w/master
On 2/28/24 06:27, John Thornton wrote: That's probably the OS theme doing that. I ran into that with Debian a few years ago the theme updated then windows not in focus became very dim. It was so annoying I searched for a new theme that didn't do that. Do you recall which one fixed this? Thanks John. Stay well. JT On 2/27/24 19:12, gene heskett wrote: Hey guys; Genuine show stopper here at the Heskett den; Running an axis gui. linuxcnc master updated yesterday. Camera gone again. Has been gone for months. Minor, this project won't use it anyway. I've almost given up on ever having a dependable camera. Some one since the last time I wrote some gcode on my g0704, has hooked the hal meter up to focus status, so if it not focused, its so dim I can't read it unless I'm under a foot from its 21" monitor, so when I click on axis so the keyboard can move he machine to manually verify the probe is working, I can't read it, unfocused, its way too dim. halmeter and halscope if its affected, so they s/b be removed from focus and made to behave as if they are focused full time. I am DIW until i can see this stuff. Is there anything I can do in hal to fix this? Thanks all. Cheers, Gene Heskett, CET. ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] got usability problem w/master
On 2/28/24 05:03, andy pugh wrote: On Wed, 28 Feb 2024 at 01:14, gene heskett wrote: Some one since the last time I wrote some gcode on my g0704, has hooked the hal meter up to focus status I am not aware of any changes made to Halmeter recently. Chris has made some stylesheet changes to QtVCP but I wouldn't expect that to affect Halmeter. Have you made any local changes to your GUI theme? Not in the last year that apt has not done in the process of keeping the buster up to date. . This is the 1 liner I use for that: sudo apt autoremove -y && sudo apt update && sudo apt list --upgradable && sudo apt upgrade -y Is halmeter actually built from a .comp? Maybe I could build it w/o that? I often have as many as 8 open while developing code so I can see whats going on in real time and now I can't see things w/o clicking on them, which means I have to click back on the axis screen before I can stop the machine with an F2 key. Dangerous to the tools. I just checked the sixty40, and the text contents are still displayed normally although the title bar is blacked out, so this problem has not affected that machine. All 3 wintels use the same 1 liner to get updated every 2 or 3 days. So I checked TLM. it is affected the text is greyed out, when not "in focus". So is the Sheldon lathe running Rod Websters arm64 build, on an rpi4b, so 3 out of 4 machines are affected. If this is a theme problem how do I go about querying the Sixty40 for whatever its running so I can install it om the other two wintels? The rpi4b is running bookworm. Untouched since Rod sent me his bookworm image. Themes are not something I fiddle with often enough to recall how. Thanks Andy. Take care and stay well. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
[Emc-developers] got usability problem w/master
Hey guys; Genuine show stopper here at the Heskett den; Running an axis gui. linuxcnc master updated yesterday. Camera gone again. Has been gone for months. Minor, this project won't use it anyway. I've almost given up on ever having a dependable camera. Some one since the last time I wrote some gcode on my g0704, has hooked the hal meter up to focus status, so if it not focused, its so dim I can't read it unless I'm under a foot from its 21" monitor, so when I click on axis so the keyboard can move he machine to manually verify the probe is working, I can't read it, unfocused, its way too dim. halmeter and halscope if its affected, so they s/b be removed from focus and made to behave as if they are focused full time. I am DIW until i can see this stuff. Is there anything I can do in hal to fix this? Thanks all. Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
[Emc-developers] linuxcnc -l is broken here
_tkinter.TclError: no display name and no $DISPLAY environment variable Shutting down and cleaning up LinuxCNC... Running HAL shutdown script task: 0 cycles, min=0.05, max=0.010078, avg=-1.00, 0 latency excursions (> 10x expected cycle time of 0.01s) hm2_5i25.0: dropping AnyIO board at :01:02.0 hm2/hm2_5i25.0: unregistered RTAPI_PCI: Unmapped 65536 bytes at 0x7fb9e6779000 hm2_pci: driver unloaded hm2: unloading Note: Using POSIX realtime LinuxCNC terminated with an error. You can find more information in the log: /home/gene/linuxcnc_debug.txt and /home/gene/linuxcnc_print.txt as well as in the output of the shell command 'dmesg' and in the terminal gene@GO704:~/linuxcnc/configs/go704-xyza$ That is a buster machine. So is tlm, and sixty40. tlm, much simpler machine and not capable of running ATM as I've only half installed the power control switching stuff so it has no motor power, but it succesfully updated and the gui still runs from a login from here in the house. Sixty40 is a much more complex machine, still runs a several day old buster update. Is this fixable or is it finally time to install bookworm?, which has NOT been a resounding success on this main machine. Plz advise as I have a job for the go704 pending. Favor for a friend on a when I get a round tuit, but... Cheers, Gene Heskett, CET. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] RPImager problem Re: RPi5 + Raspbian +LinuxCNC latency tests - first impressions
pointer, I checked and found it installed but it does not find any input or output devices. Upstream have resolved the rpi-imager compatibility issue. Their sound guy is away for a few days but has it on his to do list. The willingness to sort this out has been amazing I have to find time to merge the changes and rebuild it. I Am flat out for a while. Rod Webster *1300 896 832* +61 435 765 611 Vehicle Modifications Network www.vehiclemods.net.au On Wed, 3 Jan 2024 at 22:27, Steffen Möller via Emc-developers < emc-developers@lists.sourceforge.net> wrote: On Tue, 2 Jan 2024 01:49:34 +0100 Steffen Möller via Emc-developers wrote: The regular raspbian OS comes with sound via HDMI. I am toying around with packages to install to see if I am getting anywhere without much configuration. KDE works after installing task-kde-desktop - no sound, though. Install the pavucontrol package. That hass the controls you need to move the sound between ports. Thank you for that pointer, I checked and found it installed but it does not find any input or output devices. Best, Steffen ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] RPImager problem Re: RPi5 + Raspbian +LinuxCNC latency tests - first impressions
On 1/3/24 16:09, Chuck Peplinski wrote: If the R-Pi guys are being responsive, I wonder if you find any answer about the SPI interface for the Mesa boards? I found that linux-cnc reported SPI errors when trying to talk to the 7C81. I'm new to the Linux-CNC community, so maybe I'm doing something else wrong. But I got the idea on the forums that there was some known issues with SPI on the R-Pi5. As a user using rpspi to a mea 7i90HD, I was amazed that when subbing an rpi4b for a 3b required no changes to the spi bus. Part of the timing that makes it work is the nanosecond lags between the clock and data tines, extremely critical when the speeds are as high as they are, the write from the pi to the mesa card is 42+ megabaud! And the corresponding read is at half the 50 megacycle clocking speed of the mesa card or 25 megabaud. the rise and fall times of the actual transistors used are extremely critical and any change in the device geometry that changes this waveform cannot be well seen with a 100 mhz scope, one of the reasons I now own the fastest sampler Siglent makes, a 4 channel color channel with a 350 mhz bandwidth on all 4 channels, worth $3400 USD when I bought it nearly 3 years ago. The differences you can see are amazing. One of the things I noted was that to make it work error free on the pi3 was that it worked well with the scope probe attached to the clock but not so well w/o it, so I wound up with a 10 pf cap to ground on the pi3, but didn't need it when the pi4b was substituted. Based on that I concluded that the speed of the transistor was part of the magic sauce that made it work. Your remark makes me wonder if this same effect might be checked for with a 10 pf cap on one or the other signals to effect a few picoseconds delay difference between the clock and data lines and make it work. There is I think a parity checksum as an error detector in the protocol, with a parity error being corrected by a retransmit so if you see a sudden burst of traffic its correcting such an error.. But PCW knows that code 50,000% better than I so he is welcome to correct me. I'm just guessing based on what I see on the scope screen based on 74 years of using a scope to see what the electrical stuff is doing. 74 years ago realtime was a 5" crt with a 5 megahertz bandwidth. A Hickock 505. Sync-able but not triggered, that was a game changer. It's been on my list to follow the build instructions and see if I could find out more, but that hasn't happened yet. Thanks, Chuck On 1/3/2024 1:39 PM, Rod Webster wrote: Thank you for that pointer, I checked and found it installed but it does not find any input or output devices. Upstream have resolved the rpi-imager compatibility issue. Their sound guy is away for a few days but has it on his to do list. The willingness to sort this out has been amazing I have to find time to merge the changes and rebuild it. I Am flat out for a while. Rod Webster *1300 896 832* +61 435 765 611 Vehicle Modifications Network www.vehiclemods.net.au On Wed, 3 Jan 2024 at 22:27, Steffen Möller via Emc-developers < emc-developers@lists.sourceforge.net> wrote: On Tue, 2 Jan 2024 01:49:34 +0100 Steffen Möller via Emc-developers wrote: The regular raspbian OS comes with sound via HDMI. I am toying around with packages to install to see if I am getting anywhere without much configuration. KDE works after installing task-kde-desktop - no sound, though. Install the pavucontrol package. That hass the controls you need to move the sound between ports. Thank you for that pointer, I checked and found it installed but it does not find any input or output devices. Best, Steffen ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] RPi5 + Raspbian + LinuxCNC latency tests -first impressions
On 12/28/23 18:03, andy pugh wrote: On Thu, 28 Dec 2023 at 14:07, Daniel Hiepler wrote:oot I had no more Wifi and (worse) the ventilator was a max I encountered similar problems and decided to build my own kernel. I created a "fire & forget" bash script to build an image from the ground up and compiles linuxcnc+rt kernel on the pi: Why do people keep reinventing this particular wheel? We distribute a Pi SD card image via www.linuxcnc.org. If there is something about it that you don't like, then make a change request, rather than dividing the effort. Not exactly Andy, 2.9.2 will probably run my stuff just fine but I enjoy running the bleeding edge, makes me feel like I'm contributing in some small way to the quality of the code going out the door, basically playing the part of the canary in a cage in a coal mine. But generally, you guys are so good that in 20 years I've only reported 3 gotcha's. That, compared to what Redmond does, is legendary and every one of you should be proud it. Take care, all of you, stay warm and well. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] RPi5 + Raspbian + LinuxCNClatencytests-firstimpressions
On 12/28/23 15:34, Rod Webster wrote: The old buildbot ran on a pi but it was not robust and often pulled the whole system down. Mine is the exact opposite. With a good psu. they run from forced reboot to forced reboot, A year at a time or more, ditto the bananpi-m5's. I haven't tried on the bpi's but my std power supply for any pi/clone pi has 5 amps if it needs it. The wallwarts, not so good. Rob M built a cross compiled buildbot solution using what we use for the official images but there was no response. To fix the buildbot somebody in charge of it actually has to want to fix it. Its really too late to offer solutions now, get behind the current build process and work with it. Rod Webster *1300 896 832* +61 435 765 611 Vehicle Modifications Network www.vehiclemods.net.au On Fri, 29 Dec 2023 at 03:04, gene heskett wrote: On 12/28/23 09:05, Andy Pugh wrote: On 28 Dec 2023, at 12:55, gene heskett wrote: Any chance of updating the buildbots page yellow table? I can’t. I have no access to the buildbot. Anyway, the buildbot doesn’t build anything for arm64. Is that a lack of hardware? I have offered some in the past and been ignored, tax laws or whatever, IDK and don't really care Andy, and I can afford it w/o shortening my ladder up the side of the hog. TANSTAAFL is a law you can't break even if you try. I've been an occasional PITA here for 20 years, for free. Maybe its time I pay the rent with a charitable hardware donation? Take care, stay warm and well and have a better 2024. ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] RPi5 + Raspbian + LinuxCNC latencytests-firstimpressions
On 12/28/23 09:05, Andy Pugh wrote: On 28 Dec 2023, at 12:55, gene heskett wrote: Any chance of updating the buildbots page yellow table? I can’t. I have no access to the buildbot. Anyway, the buildbot doesn’t build anything for arm64. Is that a lack of hardware? I have offered some in the past and been ignored, tax laws or whatever, IDK and don't really care Andy, and I can afford it w/o shortening my ladder up the side of the hog. TANSTAAFL is a law you can't break even if you try. I've been an occasional PITA here for 20 years, for free. Maybe its time I pay the rent with a charitable hardware donation? Take care, stay warm and well and have a better 2024. ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] RPi5 + Raspbian + LinuxCNC latency tests-firstimpressions
On 12/28/23 06:46, andy pugh wrote: On Thu, 28 Dec 2023 at 03:24, gene heskett wrote: All packages are up to date. W: Skipping acquire of configured file '2.9-rt/source/Sources' as repository 'https://www.linuxcnc.org bookworm InRelease' does not seem to provide it (sources.list entry misspelt?) There isn't a 2.9-rt for Pi, only a 2.9-uspace. You can remove 2.9-rt from the sources.list.d if that message bothers you. Any chance of updating the buildbots page yellow table?, I'd druther get the daily builds of master in uspace mode. Thanks Andy. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] RPi5 + Raspbian + LinuxCNC latency tests-firstimpressions
On 12/28/23 02:34, Rod Webster wrote: The Pi image is forked from here https://github.com/pyavitz/rpi-img-builder It is based on the official Raspberry Pi kernel https://github.com/raspberrypi/linux (rpi-6.1-y branch ) with Debian Bookworm wrapped around it. It uses a cross compilation environment so it gets built on x86 hardware. It has a good scripting environment that allows us to install linuxcnc as part of the build process.. The download page desperately needs a paragraph outlining the username (cnc) and password (cnc) and the need to run sudo menu-config to complete the build. I have no idea where the source is for that page or I would do a PR. Builds for both the pi4b/400 and the pi5 are supported by setting the board parameter at build time. The Pi4 version supports the gpiod driver which is documented. https://linuxcnc.org/docs/stable/html/drivers/hal_gpio.html Humm, apparently rpspi.ko is still part of linuxcnc. My lathe is running just fine but the gpio driver it is running on to drive a mesa 7i90HD is not this new one but the rpspi.ko that Bertho Stultans wrote back in 2017 or 2018, then rewrote in 2020 for the rpi4b when I got an rpi4b. And he was nice enough 4 yeas ago, and could probably update it for rpi5's if asked and has the data for the version in the rpi5. Running it from my buster config, it runs well and claims its master 2.10-pre which would be an armhf build. But its ONLY my config copied back into the sdcard after Rod's image was written to it. Its the linuxcnc Rod put in the image. So, what do I need to do to put the the /etc/apt/sources.list.d/linuxcnc.list for linuxcnc back on the arm64 master build version AND get and install the proper keyfiles so apt will accept it? The install instructs stops coming forward at buster. Definitely out of date now. Preempt_rt patches are supported but the challenge here is always to match the kernel being built with the patched RT version. We can specify the commit to pull so we can match to the RT patch but currently, the latest kernels that have been patched have not been merged into the pi source. As far as spi and gpio on the pi5 go, that board was never in scope. It is up to the devs to devise appropriate drivers as its out of scope of the installer. Expected. Rod Webster *1300 896 832* +61 435 765 611 Vehicle Modifications Network www.vehiclemods.net.au Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] RPi5 + Raspbian + LinuxCNC latency tests-firstimpressions
On 12/27/23 23:44, gene heskett wrote: On 12/27/23 22:33, Rob M wrote: Gene are you after a PC image or a Pi image ? rpi4 here. Or have you made a type with the amd64 bit ? No, arm64. I must admit with the "role your image utility" rod uses the locale things is a bit messy & seems to do what it wants. I've made images usingthe same method and it's felt a bit "hit & miss" getting the locale thing correct. Half the issue is a that you expect it to work like an RPi image but it's not based on any of the RPi images by "The Foundation", tho the kernel and some utilities come from there there does seem to be some "magic sauce" missing. Yes they are very against anything realtime. Mention it the 3rd time and get banned from thier forum for life. So when I first did this in about late 2017 on an rpi3b, I built my own 4.19-rt kernel but had no clue how to install it. So I pulled a tarball together that was only the lib and /boot stuff from those 2 directories of the build, which I actually did on the pi. The plain, no gzip tarball is about 28 megs, and I installed it by mounting the sd-card in a reader, and using mc, copied the lib subdir of the tarball to the /lib of the card, and the same with the /boot stuff. Put the card back in the pi's socket, it booted my kernel & ran that way till the rpi4 came out, 2 years or so. Since I do my building on an SSD, much faster than the SD card and it mounted to the rpi4 with a usb to sata cable, installing on the rpi4 was essentially a wash, rinse and repeat but much faster with the 4's usb 3 sockets. That was in Feb 2020. Its been running an updated master of linuxcnc since. And doing stuff it could not do when assembled in Chicago in the late '40's. But the image now on the card has the first 2.9 release and I'd like to go back to master & continue with what I call the canary in the coal mine duties for linuxcnc. When I get that done, trixie will probably be out, and the buster on my 3 wintel machines is getting stale so I'll bring them up to trixie, if I don't miss roll call first, a possibility since my next b-day will be the 90th. Got hardware keeping me alive now, pacemaker and a TAVR valve in my heart. 5 or 6 years left on each ;o)> So I goto linuxcnx.org & hunt up the 291 announcement, and the first thing I see when I find the build bot link is to install its key which is different foom the other one. But on execing the get key phrase, I'm greeted with: cnc@rpi4:/etc$ sudo apt-key adv --keyserver hkp://keys.openpgp.org --recv-key EF1B07FEE0EE663E Warning: apt-key is deprecated. Manage keyring files in trusted.gpg.d instead (see apt-key(8)). Executing: /tmp/apt-key-gpghome.xOf7G6JcDy/gpg.1.sh --keyserver hkp://keys.openpgp.org --recv-key EF1B07FEE0EE663E gpg: key EF1B07FEE0EE663E: new key but contains no user ID - skipped gpg: Total number processed: 1 gpg: w/o user IDs: 1 Reading the above its been deprecated, the man page agrees but offers NDA while the other man pages still say to use it. but no one has told the other apt man 8 pages. So how is this done now? On Thu, Dec 28, 2023 at 1:53 PM gene heskett wrote: On 12/27/23 20:18, andy pugh wrote: On Thu, 28 Dec 2023 at 01:15, andy pugh wrote: http://linuxcnc.org/docs/stable/html/getting-started/getting-linuxcnc.html#_raspberry_pi_image Sorry, the link to the image is higher up the same page: http://linuxcnc.org/docs/stable/html/getting-started/getting-linuxcnc.html#_normal_download Sorry, I went there and not wanting to waste the pi's card, I installed zsync on this machine too, then copy/pasted the download line, but it quickly failed: gene@coyote:~$ zsync https://www.linuxcnc.org/iso/linuxcnc_2.9.1-amd64.hybrid.iso failed on url https://www.linuxcnc.org/iso/linuxcnc_2.9.1-amd64.hybrid.iso could not read control file from URL https://www.linuxcnc.org/iso/linuxcnc_2.9.1-amd64.hybrid.iso What do I do next? Thanks all Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis _
Re: [Emc-developers] RPi5 + Raspbian + LinuxCNC latency tests-first impressions
On 12/27/23 22:33, Rob M wrote: Gene are you after a PC image or a Pi image ? rpi4 here. Or have you made a type with the amd64 bit ? No, arm64. I must admit with the "role your image utility" rod uses the locale things is a bit messy & seems to do what it wants. I've made images usingthe same method and it's felt a bit "hit & miss" getting the locale thing correct. Half the issue is a that you expect it to work like an RPi image but it's not based on any of the RPi images by "The Foundation", tho the kernel and some utilities come from there there does seem to be some "magic sauce" missing. Yes they are very against anything realtime. Mention it the 3rd time and get banned from thier forum for life. So when I first did this in about late 2017 on an rpi3b, I built my own 4.19-rt kernel but had no clue how to install it. So I pulled a tarball together that was only the lib and /boot stuff from those 2 directories of the build, which I actually did on the pi. The plain, no gzip tarball is about 28 megs, and I installed it by mounting the sd-card in a reader, and using mc, copied the lib subdir of the tarball to the /lib of the card, and the same with the /boot stuff. Put the card back in the pi's socket, it booted my kernel & ran that way till the rpi4 came out, 2 years or so. Since I do my building on an SSD, much faster than the SD card and it mounted to the rpi4 with a usb to sata cable, installing on the rpi4 was essentially a wash, rinse and repeat but much faster with the 4's usb 3 sockets. That was in Feb 2020. Its been running an updated master of linuxcnc since. And doing stuff it could not do when assembled in Chicago in the late '40's. But the image now on the card has the first 2.9 release and I'd like to go back to master & continue with what I call the canary in the coal mine duties for linuxcnc. When I get that done, trixie will probably be out, and the buster on my 3 wintel machines is getting stale so I'll bring them up to trixie, if I don't miss roll call first, a possibility since my next b-day will be the 90th. Got hardware keeping me alive now, pacemaker and a TAVR valve in my heart. 5 or 6 years left on each ;o)> On Thu, Dec 28, 2023 at 1:53 PM gene heskett wrote: On 12/27/23 20:18, andy pugh wrote: On Thu, 28 Dec 2023 at 01:15, andy pugh wrote: http://linuxcnc.org/docs/stable/html/getting-started/getting-linuxcnc.html#_raspberry_pi_image Sorry, the link to the image is higher up the same page: http://linuxcnc.org/docs/stable/html/getting-started/getting-linuxcnc.html#_normal_download Sorry, I went there and not wanting to waste the pi's card, I installed zsync on this machine too, then copy/pasted the download line, but it quickly failed: gene@coyote:~$ zsync https://www.linuxcnc.org/iso/linuxcnc_2.9.1-amd64.hybrid.iso failed on url https://www.linuxcnc.org/iso/linuxcnc_2.9.1-amd64.hybrid.iso could not read control file from URL https://www.linuxcnc.org/iso/linuxcnc_2.9.1-amd64.hybrid.iso What do I do next? Thanks all Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] RPi5 + Raspbian + LinuxCNC latency tests-first impressions
On 12/27/23 21:49, Rod Webster wrote: @gene, you can probably set the locale to your environment by typing: sudo menu-config ISTR doing that at first boot, but did locales again anyway, set everything to en-UTF8. sudo apt update still finishes up with: All packages are up to date. W: Skipping acquire of configured file '2.9-rt/source/Sources' as repository 'https://www.linuxcnc.org bookworm InRelease' does not seem to provide it (sources.list entry misspelt?) Can you post yours? I've looked at mine but don't see anything obvious. Thanks a bunch. Rod Webster *1300 896 832* +61 435 765 611 Vehicle Modifications Network www.vehiclemods.net.au On Thu, 28 Dec 2023 at 12:35, gene heskett wrote: On 12/27/23 16:50, Rod Webster wrote: You could try this image that contains preempt_rt for the Pi 5 (Linuxcnc 2.9.1) https://drive.google.com/file/d/1H_q0Ra-27ZDOhquvq179beUZQNVlex1f/view?usp=drive_link I don't have a Pi5 to test it but others say it worked. This is an updated image from our Raspberry Pi 4 one that is built for the newer Pi 5 CPU sudo apt upgrade should update to 2.9.2 I tried that with my rpi4 that is running your image, but I get this: W: GPG error: http://linuxcnc.org bookworm InRelease: The following signatures couldn't be verified because the public key is not available: NO_PUBKEY 3CB9FD148F374FEF E: The repository 'http://linuxcnc.org bookworm InRelease' is not signed. N: Updating from such a repository can't be done securely, and is therefore disabled by default. N: See apt-secure(8) manpage for repository creation and user configuration details. W: Skipping acquire of configured file '2.9-rt/source/Sources' as repository 'https://www.linuxcnc.org bookworm InRelease' does not seem to provide it (sources.list entry misspelt?) I say its installed, I've done the install twice, but here is an ls -l of: cnc@rpi4:/etc/apt$ ls -l trusted.gpg.d/ total 84 -rw-r--r-- 1 root root 11861 Mar 28 2023 debian-archive-bookworm-automatic.asc -rw-r--r-- 1 root root 11873 Mar 28 2023 debian-archive-bookworm-security-automatic.asc -rw-r--r-- 1 root root 461 Mar 28 2023 debian-archive-bookworm-stable.asc -rw-r--r-- 1 root root 11861 Mar 28 2023 debian-archive-bullseye-automatic.asc -rw-r--r-- 1 root root 11873 Mar 28 2023 debian-archive-bullseye-security-automatic.asc -rw-r--r-- 1 root root 3403 Mar 28 2023 debian-archive-bullseye-stable.asc -rw-r--r-- 1 root root 11093 Mar 28 2023 debian-archive-buster-automatic.asc -rw-r--r-- 1 root root 11105 Mar 28 2023 debian-archive-buster-security-automatic.asc -rw-r--r-- 1 root root 1704 Mar 28 2023 debian-archive-buster-stable.asc which has no sign of that key, so where did the installer put it? That key 3CB9FD148F374FEF has been installed, twice. But I can't find it. uname -a sys: Linux rpi4.coyote.den 6.1.54-rt15 #1 SMP PREEMPT_RT Wed Sep 20 20:36:44 AEST 2023 aarch64 GNU/Linux Its latency-test is not as good as my old 4.19 something build but its plenty good enough to move my lathe nicely. But from the message, I found it in /usr/share/keyrings and put a copy in trusted* and now its offering me 19 updates. Which were all security related from debian but locale complained: locale: Cannot set LC_CTYPE to default locale: No such file or directory locale: Cannot set LC_MESSAGES to default locale: No such file or directory locale: Cannot set LC_ALL to default locale: No such file or directory How to fix that? I don't have an /etc/locale but I do have these: -rw-r--r-- 1 root root2996 Jul 13 14:07 locale.alias -rw-r--r-- 1 root root9431 Oct 17 04:24 locale.gen but it recommends I run locale-gen, which returns very quickly w/o errors and does not generate anything in the locale directory, so I'll delete the directory I made. Done. And after all this I'm still bumfuzzled. And can't update linuxcnc. Thanks & Happy New Year All. Rod Webster *1300 896 832* +61 435 765 611 Vehicle Modifications Network www.vehiclemods.net.au On Thu, 28 Dec 2023 at 06:46, Sebastian Kuzminsky wrote: On 12/27/23 13:38, Steffen Möller via Emc-developers wrote: Gesendet: Mittwoch, 27. Dezember 2023 um 17:13 Uhr Von: "andy pugh" An: "EMC developers" Cc: "Steffen Möller" Betreff: Re: [Emc-developers] RPi5 + Raspbian + LinuxCNC latency tests - first impressions On Wed, 27 Dec 2023 at 15:20, Steffen Möller via Emc-developers wrote: My RPi5 arrived over Xmas and I just fired it up, was offered to install LinuxCNC directly from what we offer in Debian, and then ran latency tests. Graphics (as in video but also the extra art from your X interface when ALT-tabbing through your applications) have the most effect on the latency. I/O from the SD does not seem to affect it too much. The numbers you show are _awful_ though? It doesn't look like the LinuxCNC installation has installed the correct kernel. (I seem to recall that I had to do a fair bit of fiddling to make it happen, it's a Pi thing) Rebo
Re: [Emc-developers] RPi5 + Raspbian + LinuxCNC latency tests -first impressions
On 12/27/23 20:18, andy pugh wrote: On Thu, 28 Dec 2023 at 01:15, andy pugh wrote: http://linuxcnc.org/docs/stable/html/getting-started/getting-linuxcnc.html#_raspberry_pi_image Sorry, the link to the image is higher up the same page: http://linuxcnc.org/docs/stable/html/getting-started/getting-linuxcnc.html#_normal_download Sorry, I went there and not wanting to waste the pi's card, I installed zsync on this machine too, then copy/pasted the download line, but it quickly failed: gene@coyote:~$ zsync https://www.linuxcnc.org/iso/linuxcnc_2.9.1-amd64.hybrid.iso failed on url https://www.linuxcnc.org/iso/linuxcnc_2.9.1-amd64.hybrid.iso could not read control file from URL https://www.linuxcnc.org/iso/linuxcnc_2.9.1-amd64.hybrid.iso What do I do next? Thanks all Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] RPi5 + Raspbian + LinuxCNC latency tests -first impressions
___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] lcnc bug in master?
On 10/22/23 09:28, Rene Hopf via Emc-developers wrote: On 16.10.23 00:00, gene heskett wrote: On 10/14/23 10:17, gene heskett wrote: On 10/14/23 09:13, Sam Sokolik wrote: I have actually seen this playing with axis from master on bookworm. At some point the refresh button in axis just quits working. You have to physically re-open the gcode file for the changes to show up. Then the refresh button will work for a bit. I thought it was just me... https://github.com/LinuxCNC/linuxcnc/issues/2490 could this be related? Nope, Sam, that green reload button (axis interface) has been funkity for quite some time, but seems to have gotten more like tits on a boar hog lately. At one point yesterday, I tried it, and got different code, that contained 2 lines I had removed from that file back in September when I decided to make two different programs out of it instead. This was a 30 line file, one operation in a while loop. Reloading by filename(menu->open) was the only way I could get the results I had just edited into the file, back into lcnc, the menu->reload was also getting old code. geany launched from the menu->edit. That really ought to set a flag to do the reload automatically when geany exits. I first thought geany was starting to do gedit's 54 pickup tricks but several subsequent cats of the file were correct but lcnc was reloading old code. I just verified this on my sixty40 mill too, I have to use the menu->open pulldown to get the code I just edited 30 seconds before. Cheers, Gene Heskett. Cheers, Gene Heskett. And that is still required a/o yesterdays update to master. I still have to use file->open if I want to get the gcode I just edited 20 seconds ago. My conclusion is that there is a cache someplace that is not being invalidated by the file->edit call. And only the file->open forces a re-read of the dirs contents. The time lag, is the dir sort as the instant files name is vise-screw6.ngc and scrolling to the end in one mouse motion is foiled by the need to keep the mouse motion /ON/ the slider. It would be most appreciated here if that vertical position of the mouse was frozen for as long as the button was down, then a single swipe of the mouse would move the slider to the end of the listing. Yeah, I'm a picky being and that machines nc_files directory currently has 249 entries. ;o)> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] got nother bug in master
On 10/19/23 09:48, andy pugh wrote: On Wed, 18 Oct 2023 at 11:49, gene heskett wrote: no place to spec it as an RN tool. What's an RN tool? Radius Nose? I am fairly sure that the Axis Preview knows nothing about that. (and as it only plits the tool tip centre, it is probably only a cosmetic issue) ball nose, radius nose, round nose, take your pick, all the same thing. And yes, is cosmetic. But one of the things I'd like to see is a red trail as it moves, that actually corresponds to the diameter of the tool. That would make corrective code changes a lot more obvious. For instance, I am currently carving a 2x2 by arbitrary length of hard maple into a wood workers vise screw, with a buttress thread since the vise screw is only a one sided load. Using the 1/16" round nosed tool to do the carving. To get the 7 degree load face angle I have a wedge under the spindle motor, a 1.5 hp, 24k motor. that 7 degree tilt is the side of the tool which has a 6.35mm DOC and at the final pass I'm using 5mm of that DOC. That, when the tool is dulled by use, can put enough push/flex that it breaks off about a mm of the 5mm tall tooth, and looks like hell. Sharpness of such small tools available also varies, a lot. So for the last .1mm of the cut I've had to change the increment, to 1/4 what it is for the rest of the tooth and that is slowly cleaning up the screw. I print the half nuts which I can make any size using openscad once the screw is finished. Being able to see in the backplot, what I get would be a huge help as opposed to using a stick of maple cut from a plank of straight grained wood that cost me $180. The red point/line, single pixel trace does not show me the effects of a round nosed tool. Hopefully, the tool table will eventually grow that ability to adequately describe the tool. Backplot image should then follow at some point. Maybe by 3.1 if I'm still around and haven't missed roll call? Thanks Andy. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
[Emc-developers] got nother bug in master
I tried to set tool #9 as a 1/16" rn cutter with a length of cut .25" in the tool table. basically to get a better idea of hw has I had to increment a var to get a good finish on the back, nomially 45 degree angle of buttress thread using the rn tool. 2 problem after I put an M6 T9 in the code, first being tool 9 not found, and secondarily, no place to spec it as an RN tool. contents of tooltbl are: T1 P1 D0.246 ;1/4 inch mill T2 P2 Z0.4911 D0.125 ;1/8 mill T3 P3 D0.135 ;5/32 drill T4 P4 D0.038 ;0.040 drill T5 P5 D0.041 ;0.045 drill T6 P6 D0.052 ;0.052 pcb component lead clearance drill T7 P7 D0.0514 ;placeholder pcb bigger component clearance drill T8 P8 D0.001 ;interconnect pattern etch T9 P9 Z6.35 D1.5875 ;1/16"rn tool in mm mode T10 P10 D0.04 ;a #58 drill T11 P11 Z-0.5043 D0.0625 ;a .0625 2 fl end mill T12 P12 D0.03125 ;added 20130917 T13 P13 D0.0015 ;added 20130918 T14 P14 X-0.01 Z-0.6095 D0.025 ;roundover bit T99 P99 X-1.6505 Y0.6823 D0.0001 ;test of camview-emc, added 20131128 T15 P15 X0 Y0 Z0 D0.15748 ;4mm emd mill T16 P16 X0 Y0 Z0 D0.15748 ;4mm tool scuse the word wrap in the T99 line, t-bird can't turn it off If I remove the DOC in the z box, then it works but the tool image seems too big. I'd post a snapshot but John has yet to pass any picture I've posted that was over 150k. I have noted that others can, but not me. So I snapshotted just the backplot to show what I mean. And why is there not a shape box to indicate an RN tool. That seems pretty basic to me. Far more trouble to dup the diameter and shape of the tools cut. But someday, maybe. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] lcnc bug in master?
On 10/14/23 10:17, gene heskett wrote: On 10/14/23 09:13, Sam Sokolik wrote: I have actually seen this playing with axis from master on bookworm. At some point the refresh button in axis just quits working. You have to physically re-open the gcode file for the changes to show up. Then the refresh button will work for a bit. I thought it was just me... Nope, Sam, that green reload button (axis interface) has been funkity for quite some time, but seems to have gotten more like tits on a boar hog lately. At one point yesterday, I tried it, and got different code, that contained 2 lines I had removed from that file back in September when I decided to make two different programs out of it instead. This was a 30 line file, one operation in a while loop. Reloading by filename(menu->open) was the only way I could get the results I had just edited into the file, back into lcnc, the menu->reload was also getting old code. geany launched from the menu->edit. That really ought to set a flag to do the reload automatically when geany exits. I first thought geany was starting to do gedit's 54 pickup tricks but several subsequent cats of the file were correct but lcnc was reloading old code. I just verified this on my sixty40 mill too, I have to use the menu->open pulldown to get the code I just edited 30 seconds before. Cheers, Gene Heskett. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] double type & stupid pet tricks
On 10/14/23 09:21, Thomas J Powderly wrote: a casual look saw a _lot_ of bits in a double so i thought ' how many 7 bit ascii chars could be encoded'? allowing short messages to be sent like this: # encode 9 chars into 64bit container str="WakeuNnow" n= 0 for c in str: n = n | (ord(c) & 0x7f) n = n << 7 print(bin(n)) 0b101011110100010110101010011101101110111000 nsav = n # get the 9 chars back n = nsav l = len(str) for byt in range(l+1): c=n&0x7f cl[l-byt] = chr(c) n=n>>7 # display the chars retrieved cl ['W', 'a', 'k', 'e', 'U', 'p', 'N', 'o', 'w', '\x00'] just for s tomp Chuckle, there is a typu above TomP, the lowercase p is missing from your original string. ;o)> Neat trick, but only 88.9% compression. Other compression methods will beat that. However, one might try it on 100 kilobytes of 7 bit ascii text, then gzip -best its output. Then compare that size to a gzip -best of the original to see if it is an improvement. Then to be amazed, make a pdf of it. The pdf will probably be smaller. ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] lcnc bug in master?
On 10/14/23 09:13, Sam Sokolik wrote: I have actually seen this playing with axis from master on bookworm. At some point the refresh button in axis just quits working. You have to physically re-open the gcode file for the changes to show up. Then the refresh button will work for a bit. I thought it was just me... Nope, Sam, that green reload button (axis interface) has been funkity for quite some time, but seems to have gotten more like tits on a boar hog lately. At one point yesterday, I tried it, and got different code, that contained 2 lines I had removed from that file back in September when I decided to make two different programs out of it instead. This was a 30 line file, one operation in a while loop. Reloading by filename(menu->open) was the only way I could get the results I had just edited into the file, back into lcnc, the menu->reload was also getting old code. geany launched from the menu->edit. That really ought to set a flag to do the reload automatically when geany exits. I first thought geany was starting to do gedit's 54 pickup tricks but several subsequent cats of the file were correct but lcnc was reloading old code. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] lcnc bug in master?
On 10/14/23 05:29, andy pugh wrote: On Sat, 14 Oct 2023 at 03:10, gene heskett wrote: What I found may not actually be linuxcnc, but in linux itself. All the indications are that its a stale cache problem. The only way to get the edited code into linuxcnc is to use the pulldown to load it by displayed name, which apparently, because it has to show the dir contents, is the only way to refresh the cache. I haven't observed this problem. Which Linux version and LinuxCNC GUI are you using? axis, on uname=Linux GO704 4.19.0-25-rt-amd64 #1 SMP PREEMPT RT Debian 4.19.289-2 (2023-08-08) x86_64, and etc/issue: Debian GNU/Linux 10 Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
[Emc-developers] lcnc bug in master?
Hi guys; Tonight I was quite aggravated, editing some gcode, seeing the edit was good before I saved, but it was not making it to linuxcnc as it was loading 2 week old code. So I did some checking to see if I could get the edited code into it. What I found may not actually be linuxcnc, but in linux itself. All the indications are that its a stale cache problem. The only way to get the edited code into linuxcnc is to use the pulldown to load it by displayed name, which apparently, because it has to show the dir contents, is the only way to refresh the cache. Can you somehow cause a cache flush and reread of the dir when the reload icon is clicked on? Lose the inode info by reloading the file by name maybe? That sould fix whats turned into a major aggravation, reloading old code. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] RFC: Reduce number of HAL pin types.
On 10/11/23 06:15, andy pugh wrote: On Wed, 11 Oct 2023 at 10:58, Nicklas SB Karlsson wrote: Have programmed for many years and some programming languages have quite hard check if data types fit together and think this in general is a good feature. And I have programmed _HAL_ for over 10 years, and so think that in this case there will probably not be many problems. One example of a useful simplification is that the pin motion.motion-type has an output type of S32. The most common components that you might want to connect this to are are "enum" and "bitslice", both of which require an unsigned input. There is also the probe input disabler, enables the probe blocking if motion.motion-type is /=5. I use a select8 module for that. S64 has more than enough bits to cover both situations. How much trouble for the compiler when the target is armhf or arm64? My lathe, running on an rpi4b seems to be ok running on Rod Websters bookworm build for arm64, the latency-test results are poorer than my original 4.19 preempt-rt armhf self built kernel, but still good enough that I've not noted any stutter in its movements. Latency went from 12 microseconds to around the 75 area. Much of that diff I suspect is due to the wider stack frame, but I've no actual timing evidence to back that up. Perhaps Rod could comment? Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Upcoming stable (12.2) and oldstable (11.8)point releases
On 10/5/23 08:01, Rod Webster wrote: Andy, Please don't stop now. It's amazing what you've got done in the last week. I've put many hours into the installers which are both complete and well tested and waiting for you! If we could release 2.9 now and get it into unstable, we will be ready to act for the 12.3 point release But more importantly with some APT pinning, a 12.2 user could still access linuxcnc-uspace from the debian repos. This would be very helpful for Raspberry Pi users because our buildbot is not building for ARM64. It would be great if Debian built it for us. The only way they can get an update now is to build from source. Which I don't mind as long as it works.. I think all we can do ATM is refer those encountering this bug, is point them to a fixed .deb set. Its unfortunate that safety does not seem to carry the same priority weight as security at Debian. I will need to update the getting linuxcnc page after 2.9 is deployed. Hope we can work around that. Yes. The build instructions are also a bit dated. For those of us trying to do a locally built copy, we have a serious need to be able o select the language built, turning off the whole documentation thing isn't the best solution, IMO of course. And many thanks to both Rod and Andy, my Sheldon is running again. Latency for uspace is worse but still good enough. Thank you. Anyway, I for one am cheering you on! Make it two. Rod Webster *1300 896 832* +61 435 765 611 Vehicle Modifications Network www.vehiclemods.net.au On Thu, 5 Oct 2023 at 21:32, Steffen Möller wrote: Gesendet: Donnerstag, 05. Oktober 2023 um 13:07 Uhr Von: "andy pugh" An: "EMC developers" Betreff: Re: [Emc-developers] Upcoming stable (12.2) and oldstable (11.8) point releases On Fri, 1 Sept 2023 at 17:58, Jonathan Wiltshire wrote: The next point releases for "bookworm" (12.2) and "bullseye" (11.8) will take place on Saturday, October 7th 2023. Processing of new uploads into the relevant queues will be frozen the preceding weekend. I just noticed the last sentence. I have been working 18 hour days to hit the wrong deadline. It's already too late. Thank you for this. There will be another one, https://wiki.debian.org/DebianReleases/PointReleases says about every two months. Best, Steffen ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] 2.9 must-fix
On 10/1/23 13:13, andy pugh wrote: On Sun, 1 Oct 2023 at 18:09, gene heskett wrote: No, this is master. It, shall we say, caught me by surprise, but now tat i see how it works, I'll get used to it. The reload button ought to work, so raise an issue about it: https://github.com/LinuxCNC/linuxcnc/issues Got to login to do that , but I've long since forgotten both the username and pw. Top that with changing isp's about a decade back so I'm unk to github. But because I am known, I can't create a new account. catch 22. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] 2.9 must-fix
On 10/1/23 12:46, andy pugh wrote: On Sun, 1 Oct 2023 at 15:02, gene heskett wrote: 2 possible heads ups. Something I ran into yesterday on my go704 mill, writing gcode to carve a hard maple stick round preparatory to threading it for a vise screw, the axis interfaces "reload" button doesn't work, With 2.9? I only care about 2.9 today. No, this is master. It, shall we say, caught me by surprise, but now tat i see how it works, I'll get used to it. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] 2.9 must-fix
On 9/30/23 21:37, Chris Morley wrote:
Andy
Thanks for pushing this along a bit.
I think Phill has a pncconf fix to go in - I'll look at it now.
Chris
From: andy pugh
Sent: September 30, 2023 2:29 PM
To: EMC developers
Subject: [Emc-developers] 2.9 must-fix
I have been working through the 2.9 must-fix tagged issues on the Github.
But that list was created in Madison, several months ago, and it is
possible that there have been other issues raised since that should be
tagged as such.
Does anyone want to suggest any?
I am vaguely hopeful of releasing 2.9 in the next couple of days. Not
that anything internally has really changed to prompt this, it's more
about the upcoming debian point release focussing my mind.
2 possible heads ups.
Something I ran into yesterday on my go704 mill, writing gcode to carve
a hard maple stick round preparatory to threading it for a vise screw,
the axis interfaces "reload" button doesn't work, I had to search thru
the file selector everytime I wanted to reload a 25 line program. Sorta
inconvenient. Check the cacheing, I got the feeling it may have been
reloading from linux cache, not the just re-saved from geany on the drive.
Idea re-enforced by whats logged to the shell that launched linuxcnc:
File "/usr/lib/python3.7/shutil.py", line 104, in copyfile
raise SameFileError("{!r} and {!r} are the same file".format(src, dst))
shutil.SameFileError: '/tmp/tmpbfvnt9bb/round-vise-screw.ngc' and
'/tmp/tmpbfvnt9bb/round-vise-screw.ngc' are the same file
Which is /not/ the file I'm editing in nc_files so the reload button is
working but why is it not reloading the same file it originally loaded?
The file I'm writing and saving in nc_files? The same file I have to
negotiate the load file requestor to actually get the actual file I'm
working on in another workspace. This change makes zero sense to me.
If you want to use /tmp for a cache, the reload button should re-read
the nc_files dir, compare the times and reload the newest one.
I just tried the edit pulldown, and that is editing the file in
nc_files, so why is it the green reload icon reloading from cache in
/tmp? Doesn't make any sense to me... If its a desired change, how do
I make it work again?
I see, load the file from storage in nc_files, edit it from the copy in
/tmp using the edit pulldown, work on the cached file in /tmp, the
reload button then works, and "Save gcode as" in the pulldown to make it
permanent by copying it back to nc_files under a new name or over the
original to make it permanent.. Added complexity but safer. That will
force a change in how I do things. This machine has 315 files in the
nc_files dir. Finding the one I want takes time.
I can see the reason for the change but it also restricts me to one
workspace. I'm not convinced that is desireable. But I'll gt used to it.
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers
.
Cheers, Gene Heskett.
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author, 1940)
If we desire respect for the law, we must first make the law respectable.
- Louis D. Brandeis
Genes Web page <http://geneslinuxbox.net:6309/>
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] wiped out lcnc on my pi
On 9/27/23 16:23, Rod Webster wrote: Gene, Great you are making progress. Your scripts won't work as you refer to 32 bit armhf and this is 64bit arm64 Also Buster uses Python 2.7 which was deprecated in Bullseye. Linuxcnc now uses Python 3.11 and 2.7 is not available. There should not be any missing dependencies to build linuxcnc as the image builds linuxcnc from source. If there is you will be able to install them once we sort out your network. please follow the docs to build the packages http://linuxcnc.org/docs/devel/html/code/building-linuxcnc.html#_building_debian_packages <http://linuxcnc.org/docs/devel/html/code/building-linuxcnc.html#_building_debian_packages> when you run the debian/configure script do it this way cd ~/linuxcnc-dev/debian ./configure uspace no-docse is once we get your network to set a fixed Ip address, you should edit /etc/network/interfaces we don't need to touch dhcpcd.conf geany and nano editors are included Something like this but check the NIC name using ip a as it won't be eth0 auto eth0 iface eth0 inet static address192.0.2.7/24 <http://192.0.2.7/24> gateway 192.0.2.254 Ahh, but it is eth0 now! To set DNS servers, add something like this which will use Google's DNS servers: dns-nameservers 8.8.8.8 8.8.4.4 I normally used nsswitch, or resolv.conf to set the search order to host,namserver resolv.con already installed. And is not a link. And nameserver is my router, dd-wrt, which is running dnsmaq, and if it isn't cached, sends the request on to my isp's nameserver. I've printed both, as this machine has no access until the net is working. Thank you. About time to do something in the microwave and call it dinner. I'm near the right coast of the USA, I assume from you sig you are an Aussie? Take care & sty well. Ref: https://wiki.debian.org/NetworkConfiguration#Configuring_the_interface_manually <https://wiki.debian.org/NetworkConfiguration#Configuring_the_interface_manually> To restart networking with the new settings sudo systemctl restart networking To set your host name to rpi4.coyote.den sudo hostnamectl set-hostname rpi4.coyote.den On my network, I let the router allocate IP addresses with DHCP and if a fixed IP is required, I add a DHCP reservation on the router for the specific MAC address. That way, there is no configuring of IP addresses on the device. I think I've covered everything Rod Webster *1300 896 832* +61 435 765 611 Vehicle Modifications Network www.vehiclemods.net.au <http://www.vehiclemods.net.au/> On Thu, 28 Sept 2023 at 04:27, gene heskett <mailto:ghesk...@shentel.net>> wrote: On 9/26/23 14:54, Rod Webster wrote: > User cnc > PW cnc > Ref: > https://github.com/rodw-au/rpi-img-builder-lcnc/blob/linuxcnc/instructions.txt <https://github.com/rodw-au/rpi-img-builder-lcnc/blob/linuxcnc/instructions.txt> <https://github.com/rodw-au/rpi-img-builder-lcnc/blob/linuxcnc/instructions.txt <https://github.com/rodw-au/rpi-img-builder-lcnc/blob/linuxcnc/instructions.txt>> > Further findings, looking at /etc/network/interfaces, I am advised that for static addresses, which this is. I should consult etc/dhcpcd.conf and man dhcpcd.conf, but neither is installed. And w/o network access, I am DITW. I am black listed off the pi forum, seems I asked too many realtime questions 7 years ago when I was first setting this up on a rpi3b, which was a bit short of the power to do it smoothly as that lathe would stop for a few milliseconds now and then waiting on orders. The foundation officially does not support realtime. So I wound up building my own preempt-rt kernal, and not know anything about u-boot, I worked out a 26 megabyte tarball, the unpacked to /boot and /lib to install that kernel, which has been happily running this rpi4 from reboot to reboot. Months between reboots at times, dead stable. Right now my workspace SSD is out and a usb reader is in its usb3 socket, so my whole linuxcnc subdir has been cp -R installed. > Rod Webster > *1300 896 832* > +61 435 765 611 > Vehicle Modifications Network > www.vehiclemods.net.au <http://www.vehiclemods.net.au> <http://www.vehiclemods.net.au/ <http://www.vehiclemods.net.au/>> > > > On Wed, 27 Sept 2023 at 04:08, gene heskett mailto:ghesk...@shentel.net> > <mailto:ghesk...@shentel.net <mailto:ghesk...@shentel.net>>> wrote: > > On 9/18/23 23:32, Rod Webster wrote: > > Gene, Use this link. > > > https://drive.google.com/drive/folders/1Bv4ceKAa0nYwDvc5eVNYFtU24PQGjKfQ <https://drive.google.com/drive/folders/1Bv4ceKAa0nYwDvc5eVNYFtU24PQGjKfQ> <https://drive.google.com/drive/folders/1Bv4ceKAa0nYwDv
Re: [Emc-developers] wiped out lcnc on my pi
On 9/19/23 11:22, gene heskett wrote: On 9/17/23 18:18, andy pugh wrote: On Sun, 17 Sept 2023 at 22:24, gene heskett wrote: Try libgpiod2 for Buster. The gpiod team have done some rather odd things with their versioning. https://packages.debian.org/buster/libs/libgpiod2 Same error, not available. It's right there! If Apt can't find it, then maybe there is a problem with you repo references? It may be right there, for debian, but raspios has its own repo's. I'll get it from debian and see if it will install. But you can just click on that page and download the deb. Cheers, Gene Heskett. I may do better than that, Rod W, has been kind enough to make me a bookworm image to try. But ATM I have another iron in the fire so it will be another couple of days to see how it works, Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] wiped out lcnc on my pi
On 9/17/23 18:18, andy pugh wrote: On Sun, 17 Sept 2023 at 22:24, gene heskett wrote: Try libgpiod2 for Buster. The gpiod team have done some rather odd things with their versioning. https://packages.debian.org/buster/libs/libgpiod2 Same error, not available. It's right there! If Apt can't find it, then maybe there is a problem with you repo references? It may be right there, for debian, but raspios has its own repo's. I'll get it from debian and see if it will install. But you can just click on that page and download the deb. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] wiped out lcnc on my pi
On 9/18/23 23:32, Rod Webster wrote: Gene, Use this link. https://drive.google.com/drive/folders/1Bv4ceKAa0nYwDvc5eVNYFtU24PQGjKfQ I rebuilt the image last night as there was an update to the PREEMPT_RT patch. Got it coming in now, thanks Rod. Got it. It will be interesting to see what debian looks like on my rpi4b. I see the kernel is 6.1.46. My armbian's are up to 6.1.50 a/o a couple days ago. I'm using bannanapi-m5's to drive 3d printers running klipper. Updates are quite rapid. Its interesting that the gui for clipper is web based, and some parts of it show empty screens on the pi's firefox, but work well when I send this machines firefox to look at bpi52:80 Many thanks Rod. Also a 2.9 ISO for AMD64 is here https://drive.google.com/drive/folders/1kW9GkNFZ7eZkfQSQPkiCHNDEXyBEfdnP There may be some missing dependencies that have snuck into the Linuxcnc source.that are not listed in the deb files on both images Rod Webster *1300 896 832* +61 435 765 611 Vehicle Modifications Network www.vehiclemods.net.au Take care & stay well. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] wiped out lcnc on my pi
On 9/18/23 01:05, Rod Webster wrote: Gene There isn't really anything official for the pi 4B yet but there is an image for Master branch I built here <https://drive.google.com/file/d/1DfgpD9K1KrYJYbJwxCEQKkdvUM-uSA4E/view?usp=sharing> This picks up Andy's latest GPIO driver and has been tested as working by a few guys on the forum... It's only a week or so out of date. If you decide to build from source, this will have all of the dependencies locked in. Rod Webster *1300 896 832* +61 435 765 611 Vehicle Modifications Network www.vehiclemods.net.au Thanks Rod. I'm in the middle of something else, putting an A axis on my go704, and it may be a few days, so I've marked this post. Take care & stay well. ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] wiped out lcnc on my pi
On 9/17/23 18:23, Rod Webster wrote: Agree with Andy https://packages.debian.org/buster/libgpiod2 Something is wrong with your sources.list or you have no internet Sources list is for raspios10, not debian. Link to the debian iso for buster armhf please, and I'll convert just to get on the same page, that is whats wrong here. I'm not and never have been running debian I'm running a full, browse the net with firefox, desktop install. I'd prefer bookworm if it wasn't so broken. Or has the last respin actually been usable? IDK. Thanks Rod. Take care & stay well. Rod Webster *1300 896 832* +61 435 765 611 Vehicle Modifications Network www.vehiclemods.net.au On Mon, 18 Sept 2023 at 08:18, andy pugh wrote: On Sun, 17 Sept 2023 at 22:24, gene heskett wrote: Try libgpiod2 for Buster. The gpiod team have done some rather odd things with their versioning. https://packages.debian.org/buster/libs/libgpiod2 Same error, not available. It's right there! If Apt can't find it, then maybe there is a problem with you repo references? But you can just click on that page and download the deb. -- atp "A motorcycle is a bicycle with a pandemonium attachment and is designed for the especial use of mechanical geniuses, daredevils and lunatics." — George Fitch, Atlanta Constitution Newspaper, 1912 ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] wiped out lcnc on my pi
On 9/17/23 18:18, andy pugh wrote: On Sun, 17 Sept 2023 at 22:24, gene heskett wrote: Try libgpiod2 for Buster. The gpiod team have done some rather odd things with their versioning. https://packages.debian.org/buster/libs/libgpiod2 Same error, not available. It's right there! If Apt can't find it, then maybe there is a problem with you repo references? But you can just click on that page and download the deb. True Andy, but that didn't fix, I need the .dev pkg but the site can't find it. pi@rpi4:/media/pi/workspace $ time ./buster-make-debs.sh uspace is accepted for compatibility, but ignored successfully configured for 'uspace-Raspbian-10'.. 5 /media/pi/workspace/buster-linuxcnc dpkg-checkbuilddeps: error: Unmet build dependencies: libgpiod-dev xvfb 7 /media/pi/workspace/buster-linuxcnc dpkg-buildpackage: info: source package linuxcnc dpkg-buildpackage: info: source version 1:2.10.0~pre0 dpkg-buildpackage: info: source distribution UNRELEASED dpkg-buildpackage: info: source changed by Andy Pugh dpkg-buildpackage: info: host architecture armhf dpkg-source --before-build . dpkg-checkbuilddeps: error: Unmet build dependencies: libgpiod-dev xvfb dpkg-buildpackage: warning: build dependencies/conflicts unsatisfied; aborting dpkg-buildpackage: warning: (Use -d flag to override.) mv: cannot stat '*.deb': No such file or directory Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] wiped out lcnc on my pi
On 9/17/23 12:34, andy pugh wrote: On Sun, 17 Sept 2023 at 11:50, gene heskett wrote: sudo apt install libgpiod-dev xvfb But they don't seem to be available for raspios10. Buster Try libgpiod2 for Buster. The gpiod team have done some rather odd things with their versioning. https://packages.debian.org/buster/libs/libgpiod2 Same error, not available. -- atp "A motorcycle is a bicycle with a pandemonium attachment and is designed for the especial use of mechanical geniuses, daredevils and lunatics." — George Fitch, Atlanta Constitution Newspaper, 1912 ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] wiped out lcnc on my pi
On 9/17/23 06:26, Steffen Möller wrote: Hi Gene, The build script does not install two packages that are required for the build, according to what the self-description of the Debian package of LinuxCNC states. This is stated in the line "dpkg-checkbuilddeps: error: Unmet build dependencies: libgpiod-dev xvfb" To help out, prior to your next run of buster-make-debs.sh, run sudo apt install libgpiod-dev xvfb which should install those missing build-deps. But they don't seem to be available for raspios10. Buster The script then apparently tries to move the resulting .deb files that are not existing because the build never started because of those missing build deps. That is a(n easily fixable) problem of the script. Best, Steffen (traveling, not sending patches :-) ) . This are Gesendet: Sonntag, 17. September 2023 um 11:28 Uhr Von: "gene heskett" An: "EMC Developers" Betreff: [Emc-developers] wiped out lcnc on my pi Trying to update my rpi4 running buster, but --fix-broken removed the executables. So I try to run my buildit script which builds the debs. that fails, broken dependencies: pi@rpi4:/media/pi/workspace $ time ./buster-make-debs.sh uspace is accepted for compatibility, but ignored successfully configured for 'uspace-Raspbian-10'.. 5 /media/pi/workspace/buster-linuxcnc dpkg-checkbuilddeps: error: Unmet build dependencies: libgpiod-dev xvfb 7 /media/pi/workspace/buster-linuxcnc dpkg-buildpackage: info: source package linuxcnc dpkg-buildpackage: info: source version 1:2.10.0~pre0 dpkg-buildpackage: info: source distribution UNRELEASED dpkg-buildpackage: info: source changed by Andy Pugh dpkg-buildpackage: info: host architecture armhf dpkg-source --before-build . dpkg-checkbuilddeps: error: Unmet build dependencies: libgpiod-dev xvfb dpkg-buildpackage: warning: build dependencies/conflicts unsatisfied; aborting dpkg-buildpackage: warning: (Use -d flag to override.) mv: cannot stat '*.deb': No such file or directory real0m4.396s user0m3.275s sys 0m0.866s What should I do next? Thank you. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers . Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] wiped out lcnc on my pi
On 9/17/23 06:26, Steffen Möller wrote: Hi Gene, The build script does not install two packages that are required for the build, according to what the self-description of the Debian package of LinuxCNC states. This is stated in the line "dpkg-checkbuilddeps: error: Unmet build dependencies: libgpiod-dev xvfb" To help out, prior to your next run of buster-make-debs.sh, run sudo apt install libgpiod-dev xvfb No such critters Steffen: pi@rpi4:/media/pi/workspace $ sudo apt install libgpiod-dev xvfb Reading package lists... Done Building dependency tree Reading state information... Done Package xvfb is not available, but is referred to by another package. This may mean that the package is missing, has been obsoleted, or is only available from another source E: Unable to locate package libgpiod-dev E: Package 'xvfb' has no installation candidate which should install those missing build-deps. The script then apparently tries to move the resulting .deb files that are not existing because the build never started because of those missing build deps. That is a(n easily fixable) problem of the script. Best, Steffen (traveling, not sending patches :-) ) . This are Gesendet: Sonntag, 17. September 2023 um 11:28 Uhr Von: "gene heskett" An: "EMC Developers" Betreff: [Emc-developers] wiped out lcnc on my pi Trying to update my rpi4 running buster, but --fix-broken removed the executables. So I try to run my buildit script which builds the debs. that fails, broken dependencies: pi@rpi4:/media/pi/workspace $ time ./buster-make-debs.sh uspace is accepted for compatibility, but ignored successfully configured for 'uspace-Raspbian-10'.. 5 /media/pi/workspace/buster-linuxcnc dpkg-checkbuilddeps: error: Unmet build dependencies: libgpiod-dev xvfb 7 /media/pi/workspace/buster-linuxcnc dpkg-buildpackage: info: source package linuxcnc dpkg-buildpackage: info: source version 1:2.10.0~pre0 dpkg-buildpackage: info: source distribution UNRELEASED dpkg-buildpackage: info: source changed by Andy Pugh dpkg-buildpackage: info: host architecture armhf dpkg-source --before-build . dpkg-checkbuilddeps: error: Unmet build dependencies: libgpiod-dev xvfb dpkg-buildpackage: warning: build dependencies/conflicts unsatisfied; aborting dpkg-buildpackage: warning: (Use -d flag to override.) mv: cannot stat '*.deb': No such file or directory real0m4.396s user0m3.275s sys 0m0.866s What should I do next? Thank you. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers . Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
[Emc-developers] wiped out lcnc on my pi
Trying to update my rpi4 running buster, but --fix-broken removed the executables. So I try to run my buildit script which builds the debs. that fails, broken dependencies: pi@rpi4:/media/pi/workspace $ time ./buster-make-debs.sh uspace is accepted for compatibility, but ignored successfully configured for 'uspace-Raspbian-10'.. 5 /media/pi/workspace/buster-linuxcnc dpkg-checkbuilddeps: error: Unmet build dependencies: libgpiod-dev xvfb 7 /media/pi/workspace/buster-linuxcnc dpkg-buildpackage: info: source package linuxcnc dpkg-buildpackage: info: source version 1:2.10.0~pre0 dpkg-buildpackage: info: source distribution UNRELEASED dpkg-buildpackage: info: source changed by Andy Pugh dpkg-buildpackage: info: host architecture armhf dpkg-source --before-build . dpkg-checkbuilddeps: error: Unmet build dependencies: libgpiod-dev xvfb dpkg-buildpackage: warning: build dependencies/conflicts unsatisfied; aborting dpkg-buildpackage: warning: (Use -d flag to override.) mv: cannot stat '*.deb': No such file or directory real0m4.396s user0m3.275s sys 0m0.866s What should I do next? Thank you. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] recent update broke display, can't run anything that needs display
On 9/6/23 16:12, Nathan Hartman wrote: On Wed, Sep 6, 2023 at 2:06 PM gene heskett wrote: Traceback (most recent call last): File "/usr/bin/hal_manualtoolchange", line 65, in app = tkinter.Tk(className="AxisToolChanger") File "/usr/lib/python3.7/tkinter/__init__.py", line 2023, in __init__ self.tk = _tkinter.create(screenName, baseName, className, interactive, wantobjects, useTk, sync, use) _tkinter.TclError: no display name and no $DISPLAY environment variable The last line quoted above might be an important clue: "no display name and no $DISPLAY environment variable" Perhaps there used to be a $DISPLAY envvar and it was clobbered somehow? Some rc scripts overwritten by recent updates maybe? Just a guess... -Nathan There is not a DISPLAY in the env on any of those buster machines. How would I go about it considering the ssh login is to a "XDG_SESSION_TYPE=tty" pasted from the screen I see here? ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
[Emc-developers] recent update broke display, can't run anything that needs display
2-12): IOPort hm2/hm2_5i25.0: IO Pin 033 (P2-13): IOPort hm2/hm2_5i25.0: registered hm2_5i25.0: initialized AnyIO board at :01:02.0 Axis Z Fault rehome Z required Axis B Fault rehome B required Traceback (most recent call last): File "/usr/bin/hal_manualtoolchange", line 65, in app = tkinter.Tk(className="AxisToolChanger") File "/usr/lib/python3.7/tkinter/__init__.py", line 2023, in __init__ self.tk = _tkinter.create(screenName, baseName, className, interactive, wantobjects, useTk, sync, use) _tkinter.TclError: no display name and no $DISPLAY environment variable tooldata_read_entry():comment for toolno 6 truncated to 39 chars: <0.052 pcb component lead clearance dril> tooldata_read_entry():comment for toolno 7 truncated to 39 chars: Axis Z Fault rehome Z required Axis B Fault rehome B required Traceback (most recent call last): File "/usr/bin/axis", line 123, in root_window = Tkinter.Tk(className="Axis") File "/usr/bin/axis", line 47, in __init__ OldTk.__init__(self, *args, **kw) File "/usr/lib/python3.7/tkinter/__init__.py", line 2023, in __init__ self.tk = _tkinter.create(screenName, baseName, className, interactive, wantobjects, useTk, sync, use) _tkinter.TclError: no display name and no $DISPLAY environment variable Shutting down and cleaning up LinuxCNC... task: 0 cycles, min=0.07, max=0.005082, avg=-1.00, 0 latency excursions (> 10x expected cycle time of 0.005000s) hm2_5i25.0: dropping AnyIO board at :01:02.0 hm2/hm2_5i25.0: unregistered RTAPI_PCI: Unmapped 65536 bytes at 0x7f0de992a000 hm2_pci: driver unloaded hm2: unloading Note: Using POSIX realtime LinuxCNC terminated with an error. You can find more information in the log: /home/gene/linuxcnc_debug.txt and /home/gene/linuxcnc_print.txt as well as in the output of the shell command 'dmesg' and in the terminal gene@sixty40:~$ That machine was working flawlessly a week ago, and has been updated once since. So its out of business now. The storm late last week finished off the ups on my pi running the lathe but pending poor weather made me unplug it last night, and tlm is down for power control mods, so I am totally out of business. Battery has two shorted cells, expected, its 7 years old. Battery ordered. What do I do now, folks? Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Raspberry Pi - Linuxcnc Bookworm installerreadyto go
On 9/3/23 01:46, Rod Webster wrote: Gene, There is a prebuilt .img here https://drive.google.com/drive/folders/1Bv4ceKAa0nYwDvc5eVNYFtU24PQGjKfQ?usp=sharing but it is missing the isolcpus settings. Building it will include them. I'll upload an update to the same folder when its built. There are also linuxcnc .debs there too. Be sure to checkout the linuxcnc branch and follow these instructions to build it (on a PC) https://github.com/rodw-au/rpi-img-builder-lcnc/blob/linuxcnc/instructions.txt This is for 2.9. If you want to build master, change/delete the git checkout statement in function2 found in /files/userscripts/uscripts I've been running master since forever, thanks for the heads up, tagged. Rod Webster *1300 896 832* +61 435 765 611 Vehicle Modifications Network www.vehiclemods.net.au On Sun, 3 Sept 2023 at 15:11, gene heskett wrote: On 9/2/23 23:43, Rod Webster wrote: Hi guys, I have forked the Raspberry Pi installer and built a linuxcnc branch https://github.com/rodw-au/rpi-img-builder-lcnc/tree/linuxcnc With a lot of help from upstream who made a few mods for me this is now final. (I hope) It downloads the official Raspberry Kernel at a commit that matches the latest 6.1 RT patch from kernel org and applies the 6.1.46 RT patch. It also adds the correct isolcpus settings in the boot environment. It installs xfce and downloads the latest Linuxcnc code, builds the debs and installs them to the menus. (Docs are not built as its already a 6.5 Gb .img). The source and build files are deleted on completion to reduce th footprint by about 1 Gb. It would be great if this could be forked into Linuxcnc. Also because its built in a chroot cross-compiler environment, it may be possible to incorporate into Sebs new buildbot, You can also copy the debas to another location in the script /files/userscript/uscripts You pi guys please give it a go and provide feedback on my repo. Thats me for one. But my pi is diw atm, the ups took a shot in last weeks bad weather. I'll see if I can rescue it tomorrow. Thanks Rod. Rod Webster *1300 896 832* +61 435 765 611 VMN® www.vmn.com.au <https://www.reviews.io/company-reviews/store/vmn1> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Raspberry Pi - Linuxcnc Bookworm installer readyto go
On 9/2/23 23:43, Rod Webster wrote: Hi guys, I have forked the Raspberry Pi installer and built a linuxcnc branch https://github.com/rodw-au/rpi-img-builder-lcnc/tree/linuxcnc With a lot of help from upstream who made a few mods for me this is now final. (I hope) It downloads the official Raspberry Kernel at a commit that matches the latest 6.1 RT patch from kernel org and applies the 6.1.46 RT patch. It also adds the correct isolcpus settings in the boot environment. It installs xfce and downloads the latest Linuxcnc code, builds the debs and installs them to the menus. (Docs are not built as its already a 6.5 Gb .img). The source and build files are deleted on completion to reduce th footprint by about 1 Gb. It would be great if this could be forked into Linuxcnc. Also because its built in a chroot cross-compiler environment, it may be possible to incorporate into Sebs new buildbot, You can also copy the debas to another location in the script /files/userscript/uscripts You pi guys please give it a go and provide feedback on my repo. Thats me for one. But my pi is diw atm, the ups took a shot in last weeks bad weather. I'll see if I can rescue it tomorrow. Thanks Rod. Rod Webster *1300 896 832* +61 435 765 611 VMN® www.vmn.com.au <https://www.reviews.io/company-reviews/store/vmn1> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] My rpi4b running my sheldon
On 8/18/23 12:34, Rod Webster wrote: Gene, there are no buildbots for Bookworm (or Bullseye for that matter). The Bookworm package is quite old (Feb 2023 I think) So I think you have no choice but to build from source so it may not matter which OS you use. The ARM64 support is really spotty and while there are a few unofficial images, you have no idea what they are or how old they are. I outlined a known working solution using bookworm http://linuxcnc.org/docs/2.9/html/getting-started/getting-linuxcnc.html Going this way actually shortens the dependency list a lot when you build from source. However the way you work, I would explore building the Raspberry pi Cross compiler on a X86 machine which outputs an image file you burn to the pi with the normal pi imager. I shared some links in an issue here https://github.com/LinuxCNC/linuxcnc/issues/2349 about this If you start with the upstream source and work through David Smiths git example, you will see how he patched for prempt_RT kernel and built Linuxcnc and ethercat via scripts. I link to the relevant file. I have done this and it works. Its just a bit old but it would be pretty easy to refresh it. I will have a go at this one day. My current project is a live build/installer for X86 machines using Bookworm and the buildbot debs. I got this working yesterday, it just needs a bit of rine tuning. And if its anything like the previous installs, easier to setup than debian. looking fwd to it. In the meantime the armbian people have a full i3 desktop spin of ubuntu jammy I might try booting tomorrow. That comes with a 6.1 kernel so there's at least a chance linuxcnc can run on it. Their jammy for bananapi-m5's works great with klipper running the printer. That and I've found a mailbox "you've got mail" alarm that uses a 12 volt battery kills the battery in a month, so I'm thinking up a solar cell convertor to run it. The solar I have already, makes about 6 volts in bright sunlight and I have some 1 farad 5 volt caps to make a rechargable battery. Might have the only solar powered mailbox in town when I get done. Small town. :o)> Rod Webster *1300 896 832* +61 435 765 611 Vehicle Modifications Network www.vehiclemods.net.au On Sat, 19 Aug 2023 at 01:48, gene heskett wrote: On 8/18/23 10:09, gene heskett wrote: Several problems? this morning. 1. Used synaptic to fix broken-packages by removing my locally built linuxcnc-v2.10.0 pacakages. 1a. debians arm64 netinstall, put on an sd card doesn't even try to boot. Apparently no u-boot or miss-configured for a raspberry pi or a bananapi-m5. 2. Linuxcnc en docs could then be installed from the buildbot but linuxcnc-uspace had missing dependencies and would not install from the buildbot. 3. Made new .debs from this mornings git pull and installed them with dpkg -l, but dpkg claimed missing dependency's and would not finish the configuration. missing dependency's are: === pi@rpi4:/media/pi/workspace $ sudo ./buster-lcnc-install.sh Selecting previously unselected package linuxcnc-uspace-dbgsym. (Reading database ... 422403 files and directories currently installed.) Preparing to unpack linuxcnc-uspace-dbgsym_2.10.0~pre0_armhf.deb ... Unpacking linuxcnc-uspace-dbgsym (1:2.10.0~pre0) ... Selecting previously unselected package linuxcnc-uspace-dev. Preparing to unpack linuxcnc-uspace-dev_2.10.0~pre0_armhf.deb ... Unpacking linuxcnc-uspace-dev (1:2.10.0~pre0) ... Selecting previously unselected package linuxcnc-uspace. Preparing to unpack linuxcnc-uspace_2.10.0~pre0_armhf.deb ... Unpacking linuxcnc-uspace (1:2.10.0~pre0) ... dpkg: dependency problems prevent configuration of linuxcnc-uspace: linuxcnc-uspace depends on python3-poppler-qt5; however: Package python3-poppler-qt5 is not installed. dpkg: error processing package linuxcnc-uspace (--install): dependency problems - leaving unconfigured dpkg: dependency problems prevent configuration of linuxcnc-uspace-dbgsym: linuxcnc-uspace-dbgsym depends on linuxcnc-uspace (= 1:2.10.0~pre0); however: Package linuxcnc-uspace is not configured yet. dpkg: error processing package linuxcnc-uspace-dbgsym (--install): dependency problems - leaving unconfigured dpkg: dependency problems prevent configuration of linuxcnc-uspace-dev: linuxcnc-uspace-dev depends on linuxcnc-uspace (= 1:2.10.0~pre0); however: Package linuxcnc-uspace is not configured yet. dpkg: error processing package linuxcnc-uspace-dev (--install): dependency problems - leaving unconfigured Processing triggers for man-db (2.8.5-2) ... Processing triggers for libc-bin (2.28-10+rpt2+rpi1+deb10u1) ... Processing triggers for gnome-menus (3.31.4-3) ... Processing triggers for desktop-file-utils (0.23-4) ... Processing triggers for mime-support (3.62) ... Processing triggers for hicolor-icon-theme (0.17-2) ... Processing triggers for shared-mime-info (1.10-1) ... Errors were encountered while processing: linuxcnc-uspac
Re: [Emc-developers] My rpi4b running my sheldon
On 8/18/23 10:09, gene heskett wrote: Several problems? this morning. 1. Used synaptic to fix broken-packages by removing my locally built linuxcnc-v2.10.0 pacakages. 1a. debians arm64 netinstall, put on an sd card doesn't even try to boot. Apparently no u-boot or miss-configured for a raspberry pi or a bananapi-m5. 2. Linuxcnc en docs could then be installed from the buildbot but linuxcnc-uspace had missing dependencies and would not install from the buildbot. 3. Made new .debs from this mornings git pull and installed them with dpkg -l, but dpkg claimed missing dependency's and would not finish the configuration. missing dependency's are: === pi@rpi4:/media/pi/workspace $ sudo ./buster-lcnc-install.sh Selecting previously unselected package linuxcnc-uspace-dbgsym. (Reading database ... 422403 files and directories currently installed.) Preparing to unpack linuxcnc-uspace-dbgsym_2.10.0~pre0_armhf.deb ... Unpacking linuxcnc-uspace-dbgsym (1:2.10.0~pre0) ... Selecting previously unselected package linuxcnc-uspace-dev. Preparing to unpack linuxcnc-uspace-dev_2.10.0~pre0_armhf.deb ... Unpacking linuxcnc-uspace-dev (1:2.10.0~pre0) ... Selecting previously unselected package linuxcnc-uspace. Preparing to unpack linuxcnc-uspace_2.10.0~pre0_armhf.deb ... Unpacking linuxcnc-uspace (1:2.10.0~pre0) ... dpkg: dependency problems prevent configuration of linuxcnc-uspace: linuxcnc-uspace depends on python3-poppler-qt5; however: Package python3-poppler-qt5 is not installed. dpkg: error processing package linuxcnc-uspace (--install): dependency problems - leaving unconfigured dpkg: dependency problems prevent configuration of linuxcnc-uspace-dbgsym: linuxcnc-uspace-dbgsym depends on linuxcnc-uspace (= 1:2.10.0~pre0); however: Package linuxcnc-uspace is not configured yet. dpkg: error processing package linuxcnc-uspace-dbgsym (--install): dependency problems - leaving unconfigured dpkg: dependency problems prevent configuration of linuxcnc-uspace-dev: linuxcnc-uspace-dev depends on linuxcnc-uspace (= 1:2.10.0~pre0); however: Package linuxcnc-uspace is not configured yet. dpkg: error processing package linuxcnc-uspace-dev (--install): dependency problems - leaving unconfigured Processing triggers for man-db (2.8.5-2) ... Processing triggers for libc-bin (2.28-10+rpt2+rpi1+deb10u1) ... Processing triggers for gnome-menus (3.31.4-3) ... Processing triggers for desktop-file-utils (0.23-4) ... Processing triggers for mime-support (3.62) ... Processing triggers for hicolor-icon-theme (0.17-2) ... Processing triggers for shared-mime-info (1.10-1) ... Errors were encountered while processing: linuxcnc-uspace linuxcnc-uspace-dbgsym linuxcnc-uspace-dev === But it loads and apparently runs ok anyway! So I don't appear to be disabled. 4. figured it was probably time to update from 32 bit (armhf) buster to bookworm, but on arrival at raspios site, find its may '23 release is still bullseye. dl the 64bit desktop anyway. Put it on sd card. I can edit the card for network stuff etc even before 1st boot, but since its bullseye, linuxcnc hasn't arrived yet. 5. is it worth my time to do that and switch cards and try to build linuxcnc on bullseye? The last time I tried that, the python3 was too new so configure bailed out. I reported that 2 or 3 times here, long ago when bullseye was new, and was ignored. So I'm stuck all alone out here in left field and everyone has gone home for the rest of the month/year/whatever. Please advise. And I just made another discovery. My apron jog speed enable pushbuttons have not beem working recently using the buildbot pkgs, but my locally built debs are working as intended. That pushbutton status and handling is all in hal, but the reporting to the screen so I'm advised of the speed change is pyvcp. When it didn't work I was stuck in slow, both for effect and on screen. That points at hal, but Beyond that, I have no info. Thanks. Cheers, Gene Heskett. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Downloading of DOCS still an issue
On 8/16/23 19:15, gene heskett wrote: On 8/16/23 09:50, Alan Condit via Emc-developers wrote: Gene, I was able to build linuxcnc on the Pine RockPro64 which is a 3399 chip. But you've not reported if rpspi.ko built (and worked would be super nice), and my 7i90HD hardware requires it. Alan Date: Tue, 15 Aug 2023 09:22:52 -0400 From: gene heskett To: emc-developers@lists.sourceforge.net Subject: Re: [Emc-developers] Downloading of DOCS still an issue Message-ID: <08e6b291-69c0-6ff3-3923-2a6c2edc6...@shentel.net> Content-Type: text/plain; charset=UTF-8; format=flowed Or are you running a 64-bit kernel? Not on the rpi4 just yet, its running armhf, but I've a luv affair with bananapi-m5's in my 3d printer stuff. Somewhat faster than a pi, all 4 usb ports are usb3, but no radio. $85/copy on amazon. If the pi dies, I'll see if I can rebuild rpspi.ko to run on the 3399 chip. That is helpful, but are you using the 7i90HD interface? That needs the rpspi.ko driver, which at the bare minimum would need the updated address of the gpio array as is likely different from the chip used in the rpi4b's. The author of that module is a Swedish prof named Bertho Stultans IIRC. He has been very helpful in the past. Past now would be half a decade ago. Thanks Alan. Take care & stay well. Cheers, Gene Heskett. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
[Emc-developers] My rpi4b running my sheldon
Several problems? this morning. 1. Used synaptic to fix broken-packages by removing my locally built linuxcnc-v2.10.0 pacakages. 1a. debians arm64 netinstall, put on an sd card doesn't even try to boot. Apparently no u-boot or miss-configured for a raspberry pi or a bananapi-m5. 2. Linuxcnc en docs could then be installed from the buildbot but linuxcnc-uspace had missing dependencies and would not install from the buildbot. 3. Made new .debs from this mornings git pull and installed them with dpkg -l, but dpkg claimed missing dependency's and would not finish the configuration. missing dependency's are: === pi@rpi4:/media/pi/workspace $ sudo ./buster-lcnc-install.sh Selecting previously unselected package linuxcnc-uspace-dbgsym. (Reading database ... 422403 files and directories currently installed.) Preparing to unpack linuxcnc-uspace-dbgsym_2.10.0~pre0_armhf.deb ... Unpacking linuxcnc-uspace-dbgsym (1:2.10.0~pre0) ... Selecting previously unselected package linuxcnc-uspace-dev. Preparing to unpack linuxcnc-uspace-dev_2.10.0~pre0_armhf.deb ... Unpacking linuxcnc-uspace-dev (1:2.10.0~pre0) ... Selecting previously unselected package linuxcnc-uspace. Preparing to unpack linuxcnc-uspace_2.10.0~pre0_armhf.deb ... Unpacking linuxcnc-uspace (1:2.10.0~pre0) ... dpkg: dependency problems prevent configuration of linuxcnc-uspace: linuxcnc-uspace depends on python3-poppler-qt5; however: Package python3-poppler-qt5 is not installed. dpkg: error processing package linuxcnc-uspace (--install): dependency problems - leaving unconfigured dpkg: dependency problems prevent configuration of linuxcnc-uspace-dbgsym: linuxcnc-uspace-dbgsym depends on linuxcnc-uspace (= 1:2.10.0~pre0); however: Package linuxcnc-uspace is not configured yet. dpkg: error processing package linuxcnc-uspace-dbgsym (--install): dependency problems - leaving unconfigured dpkg: dependency problems prevent configuration of linuxcnc-uspace-dev: linuxcnc-uspace-dev depends on linuxcnc-uspace (= 1:2.10.0~pre0); however: Package linuxcnc-uspace is not configured yet. dpkg: error processing package linuxcnc-uspace-dev (--install): dependency problems - leaving unconfigured Processing triggers for man-db (2.8.5-2) ... Processing triggers for libc-bin (2.28-10+rpt2+rpi1+deb10u1) ... Processing triggers for gnome-menus (3.31.4-3) ... Processing triggers for desktop-file-utils (0.23-4) ... Processing triggers for mime-support (3.62) ... Processing triggers for hicolor-icon-theme (0.17-2) ... Processing triggers for shared-mime-info (1.10-1) ... Errors were encountered while processing: linuxcnc-uspace linuxcnc-uspace-dbgsym linuxcnc-uspace-dev === But it loads and apparently runs ok anyway! So I don't appear to be disabled. 4. figured it was probably time to update from 32 bit (armhf) buster to bookworm, but on arrival at raspios site, find its may '23 release is still bullseye. dl the 64bit desktop anyway. Put it on sd card. I can edit the card for network stuff etc even before 1st boot, but since its bullseye, linuxcnc hasn't arrived yet. 5. is it worth my time to do that and switch cards and try to build linuxcnc on bullseye? The last time I tried that, the python3 was too new so configure bailed out. I reported that 2 or 3 times here, long ago when bullseye was new, and was ignored. So I'm stuck all alone out here in left field and everyone has gone home for the rest of the month/year/whatever. Please advise. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Downloading of DOCS still an issue
On 8/16/23 09:50, Alan Condit via Emc-developers wrote: Gene, I was able to build linuxcnc on the Pine RockPro64 which is a 3399 chip. Alan Date: Tue, 15 Aug 2023 09:22:52 -0400 From: gene heskett To: emc-developers@lists.sourceforge.net Subject: Re: [Emc-developers] Downloading of DOCS still an issue Message-ID: <08e6b291-69c0-6ff3-3923-2a6c2edc6...@shentel.net> Content-Type: text/plain; charset=UTF-8; format=flowed Or are you running a 64-bit kernel? Not on the rpi4 just yet, its running armhf, but I've a luv affair with bananapi-m5's in my 3d printer stuff. Somewhat faster than a pi, all 4 usb ports are usb3, but no radio. $85/copy on amazon. If the pi dies, I'll see if I can rebuild rpspi.ko to run on the 3399 chip. That is helpful, but are you using the 7i90HD interface? That needs the rpspi.ko driver, which at the bare minimum would need the updated address of the gpio array as is likely different from the chip used in the rpi4b's. The author of that module is a Swedish prof named Bertho Stultans IIRC. He has been very helpful in the past. Past now would be half a decade ago. Thanks Alan. Take care & stay well. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Downloading of DOCS still an issue
On 8/15/23 12:47, gene heskett wrote: On 8/15/23 09:32, gene heskett wrote: On 8/15/23 07:23, andy pugh wrote: On Mon, 14 Aug 2023 at 20:57, gene heskett wrote: because there is no longer an in-release file for buster, http://buildbot.linuxcnc.org/dists/buster/ . I can't even restore from the buildbot because apt won't access the site w/o that file. You could just grab the .deb file. http://buildbot.linuxcnc.org/dists/buster/master-rtpreempt/binary-armhf/ Got all 4 of them, will move them to the pi and sic dpkg on them, thanks Andy. Or are you running a 64-bit kernel? I have the arm64 netinstall but have not put it onto an sd card yet. And I've printed the release notes. Anything else I need to know to get bookworm actually running linuxcnc?? Thanks Andy. I've now put the debian arm64 netinstall on an sd card and plugged it into one of the bpi's. No attempt to boot was observed. Bringing the card back to my reader, it is not mountable, claims wrong filesystem etc. dmesg says:[764858.288180] sd 11:0:0:0: [sdd] 123596800 512-byte logical blocks: (63.3 GB/58.9 GiB) [764858.288613] sdd: detected capacity change from 0 to 123596800 [764858.289822] sdd: sdd1 sdd2 [764949.065169] EXT4-fs (sdd1): VFS: Can't find ext4 filesystem [764966.936416] FAT-fs (sdd1): bogus number of reserved sectors [764966.936425] FAT-fs (sdd1): Can't find a valid FAT filesystem [765008.230694] EXT4-fs (sdd2): VFS: Can't find ext4 filesystem [765992.599444] EXT4-fs (sdd2): VFS: Can't find ext4 filesystem blisterpack claims card is SDXC, what file system is that to linux? Can I make it usable? Addendum, put arm64/ubuntu/jammie on card, boots normally. Cheers, Gene Heskett. Cheers, Gene Heskett. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Downloading of DOCS still an issue
On 8/15/23 09:32, gene heskett wrote: On 8/15/23 07:23, andy pugh wrote: On Mon, 14 Aug 2023 at 20:57, gene heskett wrote: because there is no longer an in-release file for buster, http://buildbot.linuxcnc.org/dists/buster/ . I can't even restore from the buildbot because apt won't access the site w/o that file. You could just grab the .deb file. http://buildbot.linuxcnc.org/dists/buster/master-rtpreempt/binary-armhf/ Got all 4 of them, will move them to the pi and sic dpkg on them, thanks Andy. Or are you running a 64-bit kernel? I have the arm64 netinstall but have not put it onto an sd card yet. And I've printed the release notes. Anything else I need to know to get bookworm actually running linuxcnc?? Thanks Andy. I've now put the debian arm64 netinstall on an sd card and plugged it into one of the bpi's. No attempt to boot was observed. Bringing the card back to my reader, it is not mountable, claims wrong filesystem etc. dmesg says:[764858.288180] sd 11:0:0:0: [sdd] 123596800 512-byte logical blocks: (63.3 GB/58.9 GiB) [764858.288613] sdd: detected capacity change from 0 to 123596800 [764858.289822] sdd: sdd1 sdd2 [764949.065169] EXT4-fs (sdd1): VFS: Can't find ext4 filesystem [764966.936416] FAT-fs (sdd1): bogus number of reserved sectors [764966.936425] FAT-fs (sdd1): Can't find a valid FAT filesystem [765008.230694] EXT4-fs (sdd2): VFS: Can't find ext4 filesystem [765992.599444] EXT4-fs (sdd2): VFS: Can't find ext4 filesystem blisterpack claims card is SDXC, what file system is that to linux? Can I make it usable? Cheers, Gene Heskett. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Downloading of DOCS still an issue
On 8/15/23 07:23, andy pugh wrote: On Mon, 14 Aug 2023 at 20:57, gene heskett wrote: because there is no longer an in-release file for buster, http://buildbot.linuxcnc.org/dists/buster/ . I can't even restore from the buildbot because apt won't access the site w/o that file. You could just grab the .deb file. http://buildbot.linuxcnc.org/dists/buster/master-rtpreempt/binary-armhf/ Got all 4 of them, will move them to the pi and sic dpkg on them, thanks Andy. Or are you running a 64-bit kernel? I have the arm64 netinstall but have not put it onto an sd card yet. And I've printed the release notes. Anything else I need to know to get bookworm actually running linuxcnc?? Thanks Andy. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Downloading of DOCS still an issue
On 8/15/23 07:23, andy pugh wrote: On Mon, 14 Aug 2023 at 20:57, gene heskett wrote: because there is no longer an in-release file for buster, http://buildbot.linuxcnc.org/dists/buster/ . I can't even restore from the buildbot because apt won't access the site w/o that file. You could just grab the .deb file. http://buildbot.linuxcnc.org/dists/buster/master-rtpreempt/binary-armhf/ Or are you running a 64-bit kernel? Not on the rpi4 just yet, its running armhf, but I've a luv affair with bananapi-m5's in my 3d printer stuff. Somewhat faster than a pi, all 4 usb ports are usb3, but no radio. $85/copy on amazon. If the pi dies, I'll see if I can rebuild rpspi.ko to run on the 3399 chip. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
Re: [Emc-developers] Downloading of DOCS still an issue
On 8/14/23 11:20, andy pugh wrote: On Sun, 13 Aug 2023 at 14:07, gene heskett wrote: Because I have no interest in languages I don't speak and there is no way to build just the English docs. I am locked out of any upgrades as I'm running buster, Can you explain what you mean by this? Just that Andy, because I've installed the debs I locally made, the system is now in a broken state. Doing a fix broken will remove the locally compiled debs. and because there is no longer an in-release file for buster, the only thing I can do is save my work and install bookworm which will back me up to 2.9.0. I can't even restore from the buildbot because apt won't access the site w/o that file. Cheers, Gene Heskett. -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/> ___ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
