Re: [9fans] (no subject)
I'd certainly be happy to give it a good home if nobody else has claimed it. digby...@gmail.com if you want to discuss logistics off the list... Digby. On 2 April 2018 at 16:27, Steve Simonwrote: > Hi, > > I am in the Uk, and moving house. > > t have an HP T5325 Thin client which uses the > same Marvel chipset as the guruplug. > http://www.parkytowers.me.uk/thin/hp/t5325/index.shtml > > Someone got as far as getting one to net boot plan9 but > didn't (If I remember correctly) get the graphics working. > http://thread.gmane.org/gmane.os.plan9.general/72588 > > There is also an SATA interface which can be accessed with > a little soldering. > https://habrahabr.ru/post/260631/ > > > Free to anyone who wants it - I may ask for a postage > contribution if you are far from me. Be quick or it goes > into the skip. > > -Steve > >
Re: [9fans] (no subject)
I don't know if this is such common knowledge that nobody has bothered to mention it (please tell me either way), but much of that is now at http://9p.io/ http://9p.io/plan9/index.html and a newer one ? https://s3-us-west-2.amazonaws.com/belllabs-microsite-plan9/plan9.html was http://sources.cs.bell-labs.com materially different from these?
Re: [9fans] (no subject)
What's this? https://plan9.io/ On Fri, Jan 5, 2018 at 9:27 AM, Joseph Stewartwrote: > Someone (not me) should make a 9p to S3 service and put all the goodies > there. > -joe > > On Thu, Jan 4, 2018 at 1:20 PM, Steve Simon wrote: > >> I found the old addresses here: >> >> https://dnshistory.org >> >> plan9.bell-labs.com was 204.178.31.16 >> and sources.cs.bell-labs.com was 204.178.31.32 >> >> Both gone too, its not just DNS. >> >> I think it has fallen off its perch, >> it is are pine-ing for the fjords, >> it is an ex OS research group. >> >> -Steve >> >> >
Re: [9fans] (no subject)
Someone (not me) should make a 9p to S3 service and put all the goodies there. -joe On Thu, Jan 4, 2018 at 1:20 PM, Steve Simonwrote: > I found the old addresses here: > > https://dnshistory.org > > plan9.bell-labs.com was 204.178.31.16 > and sources.cs.bell-labs.com was 204.178.31.32 > > Both gone too, its not just DNS. > > I think it has fallen off its perch, > it is are pine-ing for the fjords, > it is an ex OS research group. > > -Steve > >
Re: [9fans] (no subject)
I found the old addresses here: https://dnshistory.org plan9.bell-labs.com was 204.178.31.16 and sources.cs.bell-labs.com was 204.178.31.32 Both gone too, its not just DNS. I think it has fallen off its perch, it is are pine-ing for the fjords, it is an ex OS research group. -Steve
Re: [9fans] (no subject)
I haven't tried for a few weeks but sources.cs.bell-labs.com has gone (from DNS), I don't know about the server behind it as I never kept a record of the IP address. has the labs server gone for good? (RIP). The entire cs.bell-labs.com domain is gone. And given that bell-labs.com has now outsourced their email to outlook.com, I think we can safely say the stake has well and truly been driven through the heart of this once grand institution. --lyndon
Re: [9fans] (no subject)
I've been grateful for the nightly vac jobs I've had going for the last several years of sources lately, though admittedly it was only for my tiny corner of contrib. On Wed, Aug 30, 2017 at 9:12 AM, Steve Simonwrote: > Hi, > > I haven't looked for a while, but http://plan9.bell-labs.com has gone, > and http://plan9.bell-labs.com/sources/ is broken - > /usr/web/plan9/sources.html has disappeared from the web server. > > Is there anyone left at the labs who might be able to fix at least the > web access to sources, it does look rather sad. > > https://9p.io/plan9 is still available, thank goodness. > > -Steve >
Re: [9fans] (no subject)
On Fri, 02 Dec 2016 11:44:52 GMT "Steve Simon"wrote: > Hi all, > > Anyone have some small and neat code (in C) to parse ISO8601 date/time > notation with all its glorious options? > > Its not hard to write, but it would be time consuming to test all the varients > so if anyone has some code, or knows of some in /sys/src that I have missed, > please tell me. May be this will help: ftp://ftp.ripe.net/test-traffic/ROOT/delayplots/iso8601.C
Re: [9fans] (no subject)
The ones that have bitten me have been ed, du and ls. I fixed cleanname not to restamp the terminating zero on the string if its already there which makes cleanname(.) work (for ls). du's problem is wrapped in String.h issues and I haven't dug any more. ed is exactly as you say. I assumed the problem was more extensive, berhaps its a non-problem with a couple of tiny fixes. Thanks, -Steve
Re: [9fans] (no subject)
what are the programs passing string constants? only found: ed.c:160: tfname = mktemp(/tmp/eX); all the other programs copy the template in a buffer before passing it to mktemp(). i think it would be better to just fix ed no? -- cinap
Re: [9fans] (no subject)
In my experience a VESA BIOS will sometimes report different available modes depending on the detected EDID. I have no problem believing this is true, but I'm also sure there's more to it than that. I agree. One problem is that these things are all different, almost completely undocumented, and implement non-standard modes seemingly at random. Some people involved with various OSX86[0] efforts have developed methods for probing (and in some cases, modifiying) the VESA BIOS on some cards. All kinds of strange behaviors have been observed. In qemu the emulated controller seems to support 1920x1200x{16,24,32} as well as some other modes term% aux/vga -m vesa -p warning: reading edid: VBE error 0x0100 vesa flagUlinear|Hlinear vesa sigVESA 2.0 vesa oemBochs/Plex86 VBE(C) 2003 http://savannah.nongnu.org/projects/vgabios/ 0.2 vesa vendor Bochs/Plex86 Developers vesa productBochs/Plex86 VBE Adapter vesa rev$Id: vbe.c,v 1.64 2011/07/19 18:25:05 vruppert Exp $ vesa cap 8-bit-dac vesa mem16777216 [...] vesa mode 0x187 1920x1200x16 r5g6b5 direct vesa mode 0x188 1920x1200x24 r8g8b8 direct vesa mode 0x189 1920x1200x32 x8r8g8b8 direct vesa mode 0x18a 2560x1600x16 r5g6b5 direct vesa mode 0x18b 2560x1600x24 r8g8b8 direct vesa mode 0x18c 2560x1600x32 x8r8g8b8 direct [...] you may be able to try these with your monitors as a start. I'd be interested in a survey with broader results than just dueling anecdotes I haven't made any attempt to catalogue VESA vs EDID discrepancies, but I do keep a small archive of hardware info here: http://plan9.stanleylieber.com/hardware We try to collect the sysinfo[1] output for as many systems as possible, for later reference. sl [0] http://www.osx86project.org [1] http://man.aiju.de/1/sysinfo
Re: [9fans] (no subject)
I'd love to hear about positive results here. Cards with VESA entries for non-4:3 modes are rare; I'm not sure I've ever seen one. The Pi, by contrast, drives my 16:10 high-res monitor without issues, which is the main reason it's my main Plan 9 terminal these days. Anthony signature.asc Description: Message signed with OpenPGP using GPGMail
Re: [9fans] (no subject)
I have got hold of a new 1920x1200 monitor but no card I can find seems to support it on plan9 - Even cards that claim this resolution and higher don't have the vesa mode entry. Anyone any thoughts, or suggestions of cards that do have the apropriate high resolution VESA modes? as anthony hints, vesa doesn't define non 4:3 modes in the standard. and vesa bioses don't typicall provide em. - erik
Re: [9fans] (no subject)
On Tue, Apr 15, 2014 at 5:36 PM, Anthony Sorace a...@9srv.net wrote: Cards with VESA entries for non-4:3 modes are rare; I'm not sure I've ever seen one. On Tue, Apr 15, 2014 at 5:50 PM, erik quanstrom quans...@quanstro.net wrote: and vesa bioses don't typicall provide em. Funny, I have 10+ machines that all use vesa, all widescreen. I have never seen a vesa BIOS that didn't provide widescreen modes. I'm sure they exist, but they certainly are not rare. I drive 1920x1200x32 with vesa and Cinap's realemu just fine. -- Aram Hăvărneanu
Re: [9fans] (no subject)
Funny, I have 10+ machines that all use vesa, all widescreen. I have never seen a vesa BIOS that didn't provide widescreen modes. I'm sure they exist, but they certainly are not rare. I drive 1920x1200x32 with vesa and Cinap's realemu just fine. my experience has been similar to anthony's. i have 10 machines, and none of them offer a non 4:3 mode from onboard graphics, at least when attached to a 4:3 monitor. i suspect that the modes offered may be different when different montors are attached, but i haven't proved that. what kind of video cards/onboard video are you using? - erik
Re: [9fans] (no subject)
In my experience a VESA BIOS will sometimes report different available modes depending on the detected EDID. The following cards reported the following modes when the following EDIDs were detected: --- 1.0.0: vid 03.00.00 1106/3344 11 0:f008 67108864 1:f400 16777216 VIA Technologies, Inc. VT3314 VIA/S3G UniChrome Pro IGP @{rfork n; aux/realemu; aux/vga -p} vesa flagUlinear|Hlinear|Fsnarf vesa sigVESA 3.0 vesa oemVIA P4N800 PRO 1.0 vesa vendor vesa product vesa rev vesa cap 8-bit-dac vesa mem16777216 vesa mode 0x101 640x480x8 m8 packed vesa mode 0x111 640x480x16 r5g6b5 direct vesa mode 0x112 640x480x32 x8r8g8b8 direct vesa mode 0x171 720x480x8 m8 packed vesa mode 0x173 720x480x16 r5g6b5 direct vesa mode 0x175 720x480x32 x8r8g8b8 direct vesa mode 0x1b9 720x540x8 m8 packed vesa mode 0x1ba 720x540x16 r5g6b5 direct vesa mode 0x1bb 720x540x32 x8r8g8b8 direct vesa mode 0x17c 720x576x8 m8 packed vesa mode 0x17e 720x576x16 r5g6b5 direct vesa mode 0x17f 720x576x32 x8r8g8b8 direct vesa mode 0x22e 800x480x8 m8 packed vesa mode 0x22f 800x480x16 r5g6b5 direct vesa mode 0x230 800x480x32 x8r8g8b8 direct vesa mode 0x103 800x600x8 m8 packed vesa mode 0x114 800x600x16 r5g6b5 direct vesa mode 0x115 800x600x32 x8r8g8b8 direct vesa mode 0x15c 848x480x8 m8 packed vesa mode 0x15d 848x480x16 r5g6b5 direct vesa mode 0x15f 848x480x32 x8r8g8b8 direct vesa mode 0x196 852x480x8 m8 packed vesa mode 0x197 852x480x16 r5g6b5 direct vesa mode 0x198 852x480x32 x8r8g8b8 direct vesa mode 0x222 853x480x8 m8 packed vesa mode 0x223 853x480x16 r5g6b5 direct vesa mode 0x224 853x480x32 x8r8g8b8 direct vesa mode 0x128 960x600x8 m8 packed vesa mode 0x129 960x600x16 r5g6b5 direct vesa mode 0x12a 960x600x32 x8r8g8b8 direct vesa mode 0x193 1024x512x8 m8 packed vesa mode 0x194 1024x512x16 r5g6b5 direct vesa mode 0x195 1024x512x32 x8r8g8b8 direct vesa mode 0x20b 1024x576x8 m8 packed vesa mode 0x20c 1024x576x16 r5g6b5 direct vesa mode 0x20d 1024x576x32 x8r8g8b8 direct vesa mode 0x200 1024x600x8 m8 packed vesa mode 0x201 1024x600x16 r5g6b5 direct vesa mode 0x203 1024x600x32 x8r8g8b8 direct vesa mode 0x204 1024x640x8 m8 packed vesa mode 0x205 1024x640x16 r5g6b5 direct vesa mode 0x206 1024x640x32 x8r8g8b8 direct vesa mode 0x105 1024x768x8 m8 packed vesa mode 0x117 1024x768x16 r5g6b5 direct vesa mode 0x118 1024x768x32 x8r8g8b8 direct vesa mode 0x161 1152x864x8 m8 packed vesa mode 0x163 1152x864x16 r5g6b5 direct vesa mode 0x164 1152x864x32 x8r8g8b8 direct vesa mode 0x125 1280x720x8 m8 packed vesa mode 0x126 1280x720x16 r5g6b5 direct vesa mode 0x127 1280x720x32 x8r8g8b8 direct vesa mode 0x179 1280x768x8 m8 packed vesa mode 0x17a 1280x768x16 r5g6b5 direct vesa mode 0x17b 1280x768x32 x8r8g8b8 direct vesa mode 0x1b6 1280x800x8 m8 packed vesa mode 0x1b7 1280x800x16 r5g6b5 direct vesa mode 0x1b8 1280x800x32 x8r8g8b8 direct vesa mode 0x13f 1280x960x8 m8 packed vesa mode 0x14f 1280x960x16 r5g6b5 direct vesa mode 0x16a 1280x960x32 x8r8g8b8 direct vesa mode 0x107 1280x1024x8 m8 packed vesa mode 0x11a 1280x1024x16 r5g6b5 direct vesa mode 0x11b 1280x1024x32 x8r8g8b8 direct vesa mode 0x16c 1360x768x8 m8 packed vesa mode 0x16d 1360x768x16 r5g6b5 direct vesa mode 0x16e 1360x768x32 x8r8g8b8 direct vesa mode 0x183 1366x768x8 m8 packed vesa mode 0x184 1366x768x16 r5g6b5 direct vesa mode 0x185 1366x768x32 x8r8g8b8 direct vesa mode 0x13b 1400x1050x8 m8 packed vesa mode 0x13c 1400x1050x16 r5g6b5 direct vesa mode 0x13e 1400x1050x32 x8r8g8b8 direct vesa mode 0x211 1440x900x8 m8 packed vesa mode 0x212 1440x900x16 r5g6b5 direct vesa mode 0x213 1440x900x32 x8r8g8b8 direct vesa mode 0x20e 1600x1024x8 m8 packed vesa mode 0x20f 1600x1024x16 r5g6b5 direct vesa mode 0x210 1600x1024x32 x8r8g8b8 direct vesa mode 0x120 1600x1200x8 m8 packed vesa mode 0x122 1600x1200x16 r5g6b5 direct vesa mode 0x124 1600x1200x32 x8r8g8b8 direct vesa mode 0x12b 1680x1050x8 m8 packed vesa mode 0x12c 1680x1050x16 r5g6b5 direct vesa mode 0x12d 1680x1050x32 x8r8g8b8 direct vesa mode 0x166 1920x1080x8 m8 packed vesa mode 0x167 1920x1080x16 r5g6b5 direct vesa mode
Re: [9fans] (no subject)
thanks for the confirmation! - erik
Re: [9fans] (no subject)
Hello, the result below is on one of my machine with GA-H61M-USB3H. the MB is nice with 9front. maia% @{rfork n; aux/realemu; aux/vga -p}|grep '^vesa mode' vesa mode 0x107 1280x1024x8 m8 packed vesa mode 0x11a 1280x1024x16 r5g6b5 direct vesa mode 0x11b 1280x1024x32 x8r8g8b8 direct vesa mode 0x105 1024x768x8 m8 packed vesa mode 0x117 1024x768x16 r5g6b5 direct vesa mode 0x118 1024x768x32 x8r8g8b8 direct vesa mode 0x112 640x480x32 x8r8g8b8 direct vesa mode 0x114 800x600x16 r5g6b5 direct vesa mode 0x115 800x600x32 x8r8g8b8 direct vesa mode 0x101 640x480x8 m8 packed vesa mode 0x103 800x600x8 m8 packed vesa mode 0x111 640x480x16 r5g6b5 direct vesa mode 0x17d 1920x1080x8 m8 packed vesa mode 0x17e 1920x1080x16 r5g6b5 direct vesa mode 0x17f 1920x1080x32 x8r8g8b8 direct maia% and maia% echo $vgasize 1920x1080x16 maia% I have wide screen display with DVI and is quite comfortable. NOTE that VGA connector cannot live with VESA mode on the MB! probably MB dependent. Kenji Arisawa 2014/04/16 7:42、erik quanstrom quans...@quanstro.net のメール: thanks for the confirmation! - erik
Re: [9fans] (no subject)
In my experience a VESA BIOS will sometimes report different available modes depending on the detected EDID. I have no problem believing this is true, but I'm also sure there's more to it than that. The device I'm most frustrated with is a Thinkpad, reporting on the built-in display, for example. I've seen the same on at least one other laptop with a widescreen display (some HP thing), and at least a pair of desktop graphics cards with a 16:10 monitor attached. Very frustrating. I'd be interested in a survey with broader results than just dueling anecdotes; I'd be happy to know I'd just gotten unlucky. Anthony signature.asc Description: Message signed with OpenPGP using GPGMail
Re: [9fans] (no subject)
In my experience a VESA BIOS will sometimes report different available modes depending on the detected EDID. I have no problem believing this is true, but I'm also sure there's more to it than that. I agree. One problem is that these things are all different, almost completely undocumented, and implement non-standard modes seemingly at random. Some people involved with various OSX86[0] efforts have developed methods for probing (and in some cases, modifiying) the VESA BIOS on some cards. All kinds of strange behaviors have been observed. I'd be interested in a survey with broader results than just dueling anecdotes I haven't made any attempt to catalogue VESA vs EDID discrepancies, but I do keep a small archive of hardware info here: http://plan9.stanleylieber.com/hardware We try to collect the sysinfo[1] output for as many systems as possible, for later reference. sl [0] http://www.osx86project.org [1] http://man.aiju.de/1/sysinfo
Re: [9fans] (no subject)
On Sat Mar 15 19:27:44 EDT 2014, jaketodd...@gmail.com wrote: Audio worked with hiro's drawterm and intel hda in 9front. since you mention the host's hardware, i'm a little confused. the host's hardware doesn't make any difference. it's drawterm's bridge between #A and the host's audio device that's the question. has someone done this for os-x? if so, where's the code? - erik
Re: [9fans] (no subject)
Audio worked with hiro's drawterm and intel hda in 9front.
Re: [9fans] (no subject)
It’s on the back burner for osx. Well, maybe the storage bin waiting for a sifter. -jas On Mar 15, 2014, at 6:07 PM, Steve Simon st...@quintile.net wrote: Subject Drawterm windows audio? Anyone added an audio driver to windows or osx drawterms? -Steve
Re: [9fans] (no subject)
what would be helpful, and move the discussion forward, is if someone could try to replicate this with unclean shutdowns after various file operations. i suspect that it won't repeat. but either way, it will move the discussion forward. For what it's worth, unclean shutdowns resulted in lost data for me under both fossil and hjfs. In my experience unclean shutdowns never seemed to cause problems for cwfs64x. In the case of both fossil and hjfs it is sometimes possible to repair the damage. Other times it is not. Fossil has vocal supporters, while hjfs is still marked experimental (bugs are actually getting fixed). The problem for users is that when you boot the system and you can't access your files, it gets in the way of the reason you booted the system in the first place. Persistence of this condition is unacceptable. I ran fossil on both hardware and under different virtual machines and eventually experienced file corruption on every single install. Once I found out about cwfs I switched to that and had zero problems, ever. Okay, I said to myself, this is where I'll stay. Anecdotal? You betcha. But cwfs never lost data. We can argue about who misread what messages forever. The fact is that some of us had problems with fossil and then found ways around the problems. For some of us that meant patching fossil or changing the way we used fossil. For others, it meant finding some other filesystem. Saying there is no problem changes nothing. You can debate with the Grand Canyon for hours, but when you walk off the cliff you're still going to plummet to the ground. http://img.stanleylieber.com/src/15358/img/1370274020.jpg -sl
Re: [9fans] (no subject)
it's a shell script that copies updated files from a Bell Labs system to sources.
Re: [9fans] (no subject)
erik quanstrom wrote: for me, the most important questions are - how do i set up a raid/hot spares, and - can i do this without rebooting. Of course, and right now I'm doing exactly that using a different operating system. Can I do that on Plan 9? I don't know, I'm trying to find out without much success. wikipedia. I really don't understand why are you sending me to read wikipedia. Generally, I think of myself as a decent speaker, I know how to make myself clear. It's obvious that in this case I failed. In my previous job I have worked on a file system that among other things also implements redundancy. If I implemented these things I guess I know about them without having to read wikipedia. The machine I use today for storage also runs bits of my own software. It's very easy to administer, tells me when disks are broken, I can just add disks for more storage without reboot and I can hot swap disks. Can Plan 9 do this? I don't know, I guess not? It's fine by me, I'm willing to sacrifice performance and ease of administration for an operating system I like better. I'm willing to implement myself what I need and doesn't exist yet, though I have a very hard time understanding what's missing from these very, very vague discussions. there's nothing strange about a sata device or even a raid of various devices of any type being presented with an ide programming interface. one could just as easily slap an ahci programming interface on, but either requires translation software/hardware. I agree that's nothing inherently strange, but in practice it's uncommon, at least in my experience. But then again, I'm not a hardware guy, so my experience means nothing. -- Aram Hăvărneanu
Re: [9fans] (no subject)
the motivation behind my question is that it's not clear to me that there is such a thing as pure hardware raid. if someone knows of something that implements the entire read/write path without a cpu, even with a degraded or rebuilding raid, i'd be very interested in that. Ahh, I see what you mean. I agree that probably there is no such thing as RAID purely implemented in hardware. I think for most people the hardware epithet is used to describe a black box where the user doesn't have access to it's internal gearing, be it hardware or software. A microwave is a hardware device for its users because the users don't need to care if the unit has firmware, discrete electronics or mechanical gears. Probably a better term would be to always use the term appliance, instead of this hardware/softare false dichotomy. (see wiki's raid article.) Plan 9 wiki? It mentions SiL 3112 SATA, 3114 SATA/RAID and VIA 82C686, VT8237 SATA/RAID, strangely, under IDE section. I haven't found yet relevant information about the SiL stuff, but the VIA stuff is what Adaptec calls HostRAID and Linux fakeRAID. A Happy New Year! -- Aram Hăvărneanu
Re: [9fans] (no subject)
(see wiki's raid article.) Plan 9 wiki? It mentions SiL 3112 SATA, 3114 SATA/RAID and VIA 82C686, VT8237 SATA/RAID, strangely, under IDE section. I haven't found yet relevant information about the SiL stuff, but the VIA stuff is what Adaptec calls HostRAID and Linux fakeRAID. wikipedia. there's nothing strange about a sata device or even a raid of various devices of any type being presented with an ide programming interface. one could just as easily slap an ahci programming interface on, but either requires translation software/hardware. for me, the most important questions are - how do i set up a raid/hot spares, and - can i do this without rebooting. - erik
Re: [9fans] (no subject)
On 12/30/2011 12:08 PM, erik quanstrom wrote: It might be a bit much for home use, but if I had a little bit of a budget I'd use Coraid's AoE stuff as the basis for my storage. Yeah, it's pretty overkill. I've previously worked at a storage company as a file system guy and now I have at home a nice array with ZFS on top. It works great, but I want to scale down. I want less stuff, not more. And I want to use Plan9, not Solaris. aoe doesn't require solaris, or any other operating system. you can use it directly with a plan 9 file server, as i do. - erik I don't think he was implying that one needed the other. In any case I figured I would ask -- are there any plans for a small scale AoE appliance from coraid? Didn't there used to be a single drive AoE kit long ago? What is the list's personal/home usage of AoE like? Do you guys use generic hardware and something like vblade (or those other ones people have made for linux like ggblade or whatever it is called)? -Jack
Re: [9fans] (no subject)
I don't think he was implying that one needed the other. In any case I figured I would ask -- are there any plans for a small scale AoE appliance from coraid? Didn't there used to be a single drive AoE kit long ago? there was once a single drive pata unit. What is the list's personal/home usage of AoE like? Do you guys use generic hardware and something like vblade (or those other ones people have made for linux like ggblade or whatever it is called)? http://www.quanstro.net/plan9/fs.html i also use vblade(8) on plan 9 for testing. - erik
Re: [9fans] (no subject)
aoe doesn't require solaris, or any other operating system. you can use it directly with a plan 9 file server, as i do. Of course AoE doesn't require much, my comment was in the context of Coraid's hardware appliance. I don't have much use for AoE at home. At one point I used it to network boot machines, but I only have laptops now, which have local disks because I need to use them disconnected from the network sometimes. I need a higher level protocol like 9p or venti, and I'd rather have a single Plan 9 machine with direct attached disks serving everything than a Plan 9 front end serving 9p and another machine providing AoE to it. I have way, way to many machines. Yesterday I've thrown away 5. I need less machines, not more :-). Does the Coraid applience implement RAID in hardware or does it use fs(3) or another software solution? -- Aram Hăvărneanu
Re: [9fans] (no subject)
My fileserver is running freebsd zfs. Basically one machine for nfs, venti, cifs, timemachine + sundry other services. This has worked well since 2005. Initially I used h/w raid under zfs. This was a mistake, forcibly corrected when my machine died. Now I use raidz. Since then I have replaced disks with much bigger disks without taking the machine down, upgraded the os zpool/zfs versions a couple of times (could've done without them but I prefer to run the latest stable release). Except for these events it has required hardly any maintenance (just checking vital signs like fan speed, any disk errors in weekly zpool scrub etc). One issue is FreeBSD install still doesn't install on zfs. But you can run a minimal root off a USB disk and manually setup zfs (which is pretty easy). I should probably run an AOE server on it as well! When I next replace this machine I will see if I can create a USB disk image. On Dec 30, 2011, at 12:35 PM, Aram Hăvărneanu ara...@mgk.ro wrote: aoe doesn't require solaris, or any other operating system. you can use it directly with a plan 9 file server, as i do. Of course AoE doesn't require much, my comment was in the context of Coraid's hardware appliance. I don't have much use for AoE at home. At one point I used it to network boot machines, but I only have laptops now, which have local disks because I need to use them disconnected from the network sometimes. I need a higher level protocol like 9p or venti, and I'd rather have a single Plan 9 machine with direct attached disks serving everything than a Plan 9 front end serving 9p and another machine providing AoE to it. I have way, way to many machines. Yesterday I've thrown away 5. I need less machines, not more :-). Does the Coraid applience implement RAID in hardware or does it use fs(3) or another software solution? -- Aram Hăvărneanu
Re: [9fans] (no subject)
I don't have much use for AoE at home. At one point I used it to network boot machines, but I only have laptops now, which have local disks because I need to use them disconnected from the network sometimes. I need a higher level protocol like 9p or venti, and I'd rather have a single Plan 9 machine with direct attached disks serving everything than a Plan 9 front end serving 9p and another machine providing AoE to it. I have way, way to many machines. Yesterday I've thrown away 5. I need less machines, not more :-). sure, but you haven't answered the question of how to do redundancy and recovery. aoe is a good way to isolate these functions into an appliance. Does the Coraid applience implement RAID in hardware or does it use fs(3) or another software solution? if a coraid appliance were pcie-attached rather than ethernet attached, would you still ask this question? do you think the block diagram of coraid hardware looks fundamentally different than the block diagram of a raid card? - erik
Re: [9fans] (no subject)
Could you do the latter without taking the machine down? On 30 December 2011 21:28, Bakul Shah ba...@bitblocks.com wrote: Since then I have replaced disks with much bigger disks without taking the machine down, upgraded the os zpool/zfs versions a couple of times
Re: [9fans] (no subject)
Charles Forsyth wrote: Since then I have replaced disks with much bigger disks without taking the machine down, upgraded the os zpool/zfs versions a couple of times Could you do the latter without taking the machine down? Upgrade zpool/zfs version yes, os, no. -- Aram Hăvărneanu
Re: [9fans] (no subject)
On 12/30/2011 3:05 PM, erik quanstrom wrote: Does the Coraid applience implement RAID in hardware or does it use fs(3) or another software solution? if a coraid appliance were pcie-attached rather than ethernet attached, would you still ask this question? do you think the block diagram of coraid hardware looks fundamentally different than the block diagram of a raid card? - erik I think he is trying to get you to divulge details on the software running on the coraid appliance itself. We all know it is plan 9 based (right?), so it would be interesting to know how this extra functionality of raid was added. I know these are trade secrets though so I've never asked. I would still ask that question if it were pcie attached. Curiosity will be my undoing though. -Jack
Re: [9fans] (no subject)
On Fri, 30 Dec 2011 21:34:39 GMT Charles Forsyth charles.fors...@gmail.com wrote: Could you do the latter without taking the machine down? On 30 December 2011 21:28, Bakul Shah ba...@bitblocks.com wrote: Since then I have replaced disks with much bigger disks without taking the machine down, upgraded the os zpool/zfs versions a couple of times zpool/zfs upgrade? Yes. Don't recall if I had to reboot afterwards. Resilvering to replace a disk takes a long time so not having to take the machine for hours is nice. A quick reboot is no big deal.
Re: [9fans] (no subject)
zpool/zfs upgrade? Yes. Don't recall if I had to reboot afterwards. You don't. -- Aram Hăvărneanu
Re: [9fans] (no subject)
That's upgrading the stored formats, not the in-kernel(?) software support for a particular version? On 30 December 2011 21:56, Aram Hăvărneanu ara...@mgk.ro wrote: zpool/zfs upgrade? Yes. Don't recall if I had to reboot afterwards. You don't. -- Aram Hăvărneanu
Re: [9fans] (no subject)
On 2011-12-30, at 14:41 PM, Charles Forsyth wrote: That's upgrading the stored formats, Yes. signature.asc Description: Message signed with OpenPGP using GPGMail
Re: [9fans] (no subject)
if a coraid appliance were pcie-attached rather than ethernet attached, would you still ask this question? do you think the block diagram of coraid hardware looks fundamentally different than the block diagram of a raid card? It's just curiosity. I know the appliance is Plan 9 based. If it uses an off-the-shelf RAID chip I might buy a card with that chip since it works in Plan 9. If it's fs(3) I know fs(3) is good enough for my needs. If it's something else at least I know fs(3) is not good enough and I might be tempted to write something myself. So yes, I'd ask even if it was a PCIe card instead of network appliance. the motivation behind my question is that it's not clear to me that there is such a thing as pure hardware raid. if someone knows of something that implements the entire read/write path without a cpu, even with a degraded or rebuilding raid, i'd be very interested in that. but as far as i know, there's always a processor in there on the other side of the bus. in case of aoe, the bus is ethernet and for a hardware raid card, it's usually some form of pci. (see wiki's raid article.) - erik
Re: [9fans] (no subject)
On Tue Aug 9 08:41:44 EDT 2011, st...@quintile.net wrote: Subject rfork N I want to construct a new namespace for an app, including a new root dir. I was hoping to do this with a few lines of script starting with rfork N, however this seems to be a dead end. I can still access kernel devices directly using the # escape in filenames, which includes '#s/fossil' for my file server. However I cannot get access to mount(1) so I can never access any files in that fossil. is rfork N in rc useless, and do I get a prize for noticing? (or is it a lack of vision on my behalf as usual?) don't you just want rfork Nm? - erik
Re: [9fans] (no subject)
don't you just want rfork Nm? I don't think so, that is worse, this would really mean I couldn't mount anything. I don't mind the new app adding mounts if it wants to, but I want it to start with a new root dir and children, so I need to flush out the old namespace. the problem is as soon as I do that I lost contact with the mount command and as its not a shell builtin this means I can never get back the namespace I have lost. I am stuck in no-mans-land. its easy enough to experiment: open a window type rfork N try to get ls to work. I can pass an open file descriptor across the rfork N boundry, but I still cannot mount it. The only solution I can see is to write some C code, but I was hoping somone might see a trick I missed. -Steve
Re: [9fans] (no subject)
I don't manage 1920x1080 but I do get 1600x1200x16 from my NVidia GeForce2 MX-200. This card is old but its what is in the machine and its works. I must say lots of pixels makes a very nice interface. -Steve
Re: [9fans] (no subject)
On Mon Apr 25 03:43:36 EDT 2011, st...@quintile.net wrote: I don't manage 1920x1080 but I do get 1600x1200x16 from my NVidia GeForce2 MX-200. This card is old but its what is in the machine and its works. I must say lots of pixels makes a very nice interface. i thought i responded to this. i run the same resolution in vesa mode on this ; pci 1002/9714 1002/9714 ATI Technologies Inc RS880 [Radeon HD 4290] this sucker supports vesa mode 0x1e5 1920x1440x16 r5g6b5 direct vesa mode 0x1e6 1920x1440x32 r8g8b8 direct but my monitor isn't that good. i also run with intel integrated graphics on an atom motherboard. - erik
Re: [9fans] (no subject)
so the front fell off? -- cinap ---BeginMessage--- On Mon Mar 28 08:07:44 EDT 2011, st...@quintile.net wrote: looks like your web server (quanstro.net) is down. yes, we had a big thunderstorm come by and knock out power for 4 hrs. we had a strike close enough to the house to blow up lightbulbs and shoot flames out of sockets shortly after power came back on, so i lost some ethernet interfaces, including the web/auth/ftp server's. both my lines are down, too. note to self: stop buying crappy consumer motherboards. hopefully i'll get it all replaced or patched up this afternoon. and with luck, the phone company will have things fixed as well. - erik ---End Message---
Re: [9fans] (no subject)
Erik, sounds like you need a motor generator ... flames from outlets ... wow ... maybe some opto isolators with a 2 meter air gap too :-) ron
Re: [9fans] (no subject)
On Mon Mar 28 11:20:39 EDT 2011, rminn...@gmail.com wrote: Erik, sounds like you need a motor generator ... flames from outlets ... wow ... maybe some opto isolators with a 2 meter air gap too :-) evidently the strike was closer to the neighbors. they lost all their major appliances, and a heat pump. (i thought you couldn't kill them.) fire breathing refrigerator! - erik
Re: [9fans] (no subject)
POSIX is a standard in which hardly anyone actually adheres too. AIX POSIX is not Solaris POSIX is not Linux POSIX etc. What good is a standard that isn't truthfully standardised. Alas I will say that POSIX does add quite a bit more cross platfom conformity than some other... things... but there a better solutions. While Plan 9 has some similiaritys to UNIX, it is not. Porting UNIX tools to something that is not is just not a good idea. You don't port a lawnmower engine to a Ferrari. You don't port horse tranquilizer to a rat. Should the few of you continue on this path, I would also ask you to port the ReactOS framework over to Plan9, and the Haiku kits, and AROS's librarys, and... and... A big reason for the creation of Plan 9 was to get away from the complexity, from the everything but the kitchen sink mentality. I would strongly suggest reading the articles on http://plan9.bell-labs.com/sys/doc/ before you continue your massochistic crusade. On Tue, Mar 9, 2010 at 12:50 AM, Lyndon Nerenberg (VE6BBM/VE7TFX) lyn...@orthanc.ca wrote: surely your joking, mr. nerenberg! Nope. Over the past 10 years I can only think of one or two projects I did that required platform-specific optimizations outside of POSIX.
Re: [9fans] (no subject)
Thing is, It's autoconf that needs careful redesign s/redesign/annihilation/ Yes, but even if everybody were to switch to Linux/386 as the only available platform, there would remain a niche for something like autoconf, programmers just don't change habits very quickly. The trick is to make sure the next instance isn't as monstrous, knowing that normally the next generation is more monstrous than the previous one. Nice thing is, annihilation does not have to be careful :-) ++L
Re: [9fans] (no subject)
-Original Message- From: 9fans-boun...@9fans.net [mailto:9fans-boun...@9fans.net] On Behalf Of lu...@proxima.alt.za Sent: Sunday, March 07, 2010 11:58 PM To: 9fans@9fans.net Subject: Re: [9fans] (no subject) Agreed wholeheartedly. Thing is, It's autoconf that needs careful redesign: I don't see any need for autoconf. As one wise person put it to me, things like configure and autoconf just mean you don't know how to write portable code. Again, agreed, but reality out there suggests many others are still believers, no matter how misguided. We were discussing making available Open Source ports... What your discussing is adding massive complexity. I would prefer to not have the tools and programs, and retain simplicity, rather than add outlandish complexity. I still like to point people at plan 9 ports as an example of a complex system that gets by without this *conf* nonsense. It falls over just enough to be attacked. Otherwise, p9p source would have been ported back to Plan 9 in its entirety. It's a shame, really, and with some work it could be fixed, but some of that work is design work. ++L PS: I realise that I'm proposing two nearly orthogonal objectives, sorry that I didn't clarify that sooner.
Re: [9fans] (no subject)
At lsub we are doing a bit of work on a db for linux. To make a long story short, autoconf/automake/xmkmf/whatelse? made the code so non portable that a particular version of suse is now necessary to run the thing. Isn't that the opposite of what auto* was trying to achieve? I'd just say: say no. On Mon, Mar 8, 2010 at 4:13 PM, Patrick Kelly kameo76...@gmail.com wrote: -Original Message- From: 9fans-boun...@9fans.net [mailto:9fans-boun...@9fans.net] On Behalf Of lu...@proxima.alt.za Sent: Sunday, March 07, 2010 11:58 PM To: 9fans@9fans.net Subject: Re: [9fans] (no subject) Agreed wholeheartedly. Thing is, It's autoconf that needs careful redesign: I don't see any need for autoconf. As one wise person put it to me, things like configure and autoconf just mean you don't know how to write portable code. Again, agreed, but reality out there suggests many others are still believers, no matter how misguided. We were discussing making available Open Source ports... What your discussing is adding massive complexity. I would prefer to not have the tools and programs, and retain simplicity, rather than add outlandish complexity. I still like to point people at plan 9 ports as an example of a complex system that gets by without this *conf* nonsense. It falls over just enough to be attacked. Otherwise, p9p source would have been ported back to Plan 9 in its entirety. It's a shame, really, and with some work it could be fixed, but some of that work is design work. ++L PS: I realise that I'm proposing two nearly orthogonal objectives, sorry that I didn't clarify that sooner.
Re: [9fans] (no subject)
Thing is, It's autoconf that needs careful redesign: The question is, why should I care about that all? I have mk(1) and I really don't need more to build software on Plan9 (rc, awk/sed, etc. come in handy). You seem to insist on alien software, why is porting software from lunix a prefered solution? Most of the solutions are a.) either to problems we don't even have or b.) so awkward in interfacing you just end up writing a native fork anyway. Another interesting quesiton is: Should we have the ability to port alien software to Plan9, from an ethical point of view? I personally would feel better if APE was in extra/ (and has it's own mailinglist and contrib/). even though contrib/replica is not as comprehensive as a revision control even when backed by venti, it is unmatched for Plan 9 purposes. parse error. What are unmatched Plan9 purposes? And what has autoconf to do with contrib and replica?
Re: [9fans] (no subject)
You seem to insist on alien software, why is porting software from lunix a prefered solution? Most of the solutions are a.) either to problems we don't even have or b.) so awkward in interfacing you just end up writing a native fork anyway. That's simple pragmatism. Using Plan 9 without some of the Open Source stuff is hard at least for some of us. And there just aren't enough Plan 9 developers to produce alternatives. Considering the number of useful (if poorly implemented) Open Source tools out there, I'm sure I'm not being absurd. ++L
Re: [9fans] (no subject)
And there just aren't enough Plan 9 developers to produce alternatives. Then there cannot possibly be enough to port the auto* abortion.
Re: [9fans] (no subject)
On Mon Mar 8 13:04:41 EST 2010, lu...@proxima.alt.za wrote: You seem to insist on alien software, why is porting software from lunix a prefered solution? Most of the solutions are a.) either to problems we don't even have or b.) so awkward in interfacing you just end up writing a native fork anyway. That's simple pragmatism. Using Plan 9 without some of the Open Source stuff is hard at least for some of us. And there just aren't enough Plan 9 developers to produce alternatives. Considering the number of useful (if poorly implemented) Open Source tools out there, I'm sure I'm not being absurd. regardless, this thread is turning into a trollfest. minds are made up. no one is being swayed by the arguments. (and all the arguments have already made the rounds.) i think the only way to redeem it is write some code. failing that, let's at least give this thread a proper burial. - erik
Re: [9fans] (no subject)
lu...@proxima.alt.za wrote: That's simple pragmatism. Using Plan 9 without some of the Open Source stuff is hard at least for some of us. And there just aren't enough Plan 9 developers to produce alternatives. Considering the number of useful (if poorly implemented) Open Source tools out there, I'm sure I'm not being absurd. What tools do you have in mind? This can be true for useful projects but it is not the case with Autotools, because it is properly implemented as a useless tool. If they wrote their own Makefile.in you can use that or Makefile.am to create a mkfile fairly quickly. There's no need to worry about path searching because of Plan 9 filespace unions. Testing for missing headers/libraries is also pointless as compiling will tell you this. Autotools will not generate portable code, it can only inject code that the author wrote a condition for. effort reward ergo GNU Not Useful
Re: [9fans] (no subject)
Then there cannot possibly be enough to port the auto* abortion. I hope so too, that would be insane. But there ought to be a sane alternative and it should not be anywhere as complex. I'm sorry I put the NetBSD package system in the same basket as the autoconf stuff, their relationship is merely that the package system is geared towards autoconf, but that is only a fraction of its functionality and I'm sure can be replaced. In fact, thinking about it, I've often wished the configuration stuff would move out of the individual packages so that duplication could be eliminated. So maybe that's what _I'm_ looking for. But enough said. I have a much better idea of what the vocal Plan 9 community would (not) like. ++L
Re: [9fans] (no subject)
What tools do you have in mind? This can be true for useful projects but it is not the case with Autotools, because it is properly implemented as a useless tool. No, I do not have the autotools in mind, I share the prevalent dislike for them. Just to clarify, I use dot, fgb has ported both OpenSSH and OpenSSL with good cause. Your comments about generating a mkfile from Makefile.am are the most constructive I have seen so far. ++L
Re: [9fans] (no subject)
But there ought to be a sane alternative and it should not be anywhere as complex. There is: it's called POSIX.
Re: [9fans] (no subject)
On Tue Mar 9 00:27:59 EST 2010, lyn...@orthanc.ca wrote: But there ought to be a sane alternative and it should not be anywhere as complex. There is: it's called POSIX. surely your joking, mr. nerenberg! - erik
Re: [9fans] (no subject)
surely your joking, mr. nerenberg! Nope. Over the past 10 years I can only think of one or two projects I did that required platform-specific optimizations outside of POSIX.
Re: [9fans] (no subject)
My gripe here is that it is hard to track what has been ported and what hasn't and repetition isn't helpful. grep something /n/sources/lsr ?
Re: [9fans] (no subject)
My gripe here is that it is hard to track what has been ported and what hasn't and repetition isn't helpful. grep something /n/sources/lsr ? I wasn't aware of an lsr, but I don't think that's it, really. One needs more than a file name in many instances. If I had infinite free resources at my disposal, I'd use the NetBSD packages DESCR files in some guise or other. Of course, one would then be tempted (as I have been) to look more seriously at porting the NetBSD package system to Plan 9. That's not out of the question, in fact it's probably not too difficult, but the residual pain of autoconf for each individual package has already frightened people with a stronger stomach than mine away. That is why I cannot suggest we catch up, but it may be nice to be ready when eventually the autoconf edifice falls down and something mildly intelligent takes its place! Ironically, Plan 9 may be the platform on which a replacement for autoconf could be designed and implemented. But that's too close to a pipedream to be given serious consideration. ++L
Re: [9fans] (no subject)
We have fgb's contrib, and before that just the INDEX files in / contrib on sources. Neither is a perfect solution, but I don't think the problem here would be addressed by the Labs providing some new resource. Between the above and the wiki, there's plenty of opportunity for folks to make ports known.
Re: [9fans] (no subject)
On Sat, Mar 6, 2010 at 9:05 PM, Anthony Sorace a...@9srv.net wrote: We have fgb's contrib, and before that just the INDEX files in /contrib on sources. we've got fgb's wonderful program and I think we're crazy if we don't build on that. Or we're CADT. ron
Re: [9fans] (no subject)
we've got fgb's wonderful program and I think we're crazy if we don't build on that. Or we're CADT. haven't you heard? we're not allowed to do anything for ourselves anymore. the mythical (and god like) Library Writers do this for us. our job is to glue things together and port: http://developers.slashdot.org/story/10/03/07/0043215/Whatever-Happened-To-Programming?art_pos=8art_pos=9art_pos=9 if we were to write our own, we would be guilty of Wasting Resources. this is a capital offense. therefore we must port what the Library Writers have given us. ... or maybe not. i think contrib is pretty nice. why do we need blessed packages, away? if you're installing all unix-plan 9 ports, perhaps you would be happer running unix in the first place. - erik
Re: [9fans] (no subject)
We have fgb's contrib, and before that just the INDEX files in / contrib on sources. Neither is a perfect solution, but I don't think the problem here would be addressed by the Labs providing some new resource. Between the above and the wiki, there's plenty of opportunity for folks to make ports known. I'm merely suggesting a grouping function and I certainly am not in a position to prescribe how it should be implemented. As I mentioned, I like the way NetBSD's package does it, but the price is very steep. Fgb's contrib sounds very good, I have not had occasion to try it but I presume it retains the scattered nature of the contrib directory. My choice would be to add a directory wherein to store both modified sources and binaries for Open Source projects once they have been validated. Of necessity, one would have the version clearly indicated and where possible duplications as occur frequently with popular packages such a zlib would be removed. But there seems to me to be a need to keep them together, although that may be just that I'm looking at the problem from the single perspective of how it's done in NetBSD. ++L
Re: [9fans] (no subject)
Fgb's contrib sounds very good, I have not had occasion to try it but I presume it retains the scattered nature of the contrib directory. [...] I'm looking at the problem from the single perspective of how it's done in NetBSD. really? you haven't even tried it and your trying to fit it into a whatever-netbsd-does shoebox? - erik
Re: [9fans] (no subject)
really? you haven't even tried it and your trying to fit it into a whatever-netbsd-does shoebox? Well, I do have some idea on how fgb's contrib is meant to work and, yes, I do look at it from a biased perspective :-) Surely you don't think that is implicitly flawed logic? After all, how familiar are you with the NetBSD package system? Fgb's contrib is not too different, although it is certainly more relaxed, in my opinion out of historical necessity. Again in my opinion the additional flexibility makes it harder to categorise packages and such categorisation would be of greater benefit. But it is just an opinion. ++L
Re: [9fans] (no subject)
we've got fgb's wonderful program and I think we're crazy if we don't build on that. A deserved compliment, certainly. Now to figure how to provide the grouping I believe is required in addition to the means for replication. And, if I'm not being stupid, some facility to make sure that the final product has at least some chance of working according to expectations. Can't expect fgb to deal with that :-) ++L
Re: [9fans] (no subject)
if you're installing all unix-plan 9 ports, perhaps you would be happer running unix in the first place. That's flawed logic: I may need dot, while you need curses. It's nice if both have been blessed. ++L
Re: [9fans] (no subject)
On Sun, Mar 7, 2010 at 1:09 PM, lu...@proxima.alt.za wrote: we've got fgb's wonderful program and I think we're crazy if we don't build on that. A deserved compliment, certainly. Now to figure how to provide the grouping I believe is required in addition to the means for replication. And, if I'm not being stupid, some facility to make sure that the final product has at least some chance of working according to expectations. Can't expect fgb to deal with that :-) ++L Wouldn't grouping be a matter of placing a category file in the package? I'm reading grouping as the kind of divisions you get in Ports, i.e. net, editor, util, language, etc. John -- Object-oriented design is the roman numerals of computing -- Rob Pike
Re: [9fans] (no subject)
On Sun, Mar 7, 2010 at 2:42 PM, lu...@proxima.alt.za wrote: We have fgb's contrib, and before that just the INDEX files in / contrib on sources. Neither is a perfect solution, but I don't think the problem here would be addressed by the Labs providing some new resource. Between the above and the wiki, there's plenty of opportunity for folks to make ports known. I'm merely suggesting a grouping function and I certainly am not in a position to prescribe how it should be implemented. As I mentioned, I like the way NetBSD's package does it, but the price is very steep. Fgb's contrib sounds very good, I have not had occasion to try it but I presume it retains the scattered nature of the contrib directory. My choice would be to add a directory wherein to store both modified sources and binaries for Open Source projects once they have been validated. who's gonna validate the beasts? Of necessity, one would have the version clearly indicated and where possible duplications as occur frequently with popular packages such a zlib would be removed. But there seems to me to be a need to keep them together, although that may be just that I'm looking at the problem from the single perspective of how it's done in NetBSD. please take 10 minutes to try fgb/contrib. while at it, run contrib/gui. when a package is duplicated, it should be clear from the package list on the left. iru
Re: [9fans] (no subject)
Wouldn't grouping be a matter of placing a category file in the package? I'm reading grouping as the kind of divisions you get in Ports, i.e. net, editor, util, language, etc. Thing is, the port hierarchy (hereto I used NetBSD package system for the same concept) provides both the hierarchical structure I believe is needed to minimise duplication and a description file to search for concepts rather than file names. So, yes, I agree with a portion of your suggestion, but not your suggested implementation. In practice, all I'm proposing is replacing directories owned by individual contributors with a hierarchy that undergoes a useful amount of validation. I'm sure that fgb's contrib, possibly with manageable alterations, will successfully deal with my proposal, with the reservation that we probably want a selection mechanism that reduces the magnitude of updates/replications to manageable proportions. The NetBSD packages hierarchy skeleton is gargantuan, I think we ought to provide tools to trim down any new implementation of it, but I have not thought to any extent how we'd do that. ++L
Re: [9fans] (no subject)
That's flawed logic: I may need dot, while you need curses. It's nice if both have been blessed. bless a curse! jehova! -- cinap
Re: [9fans] (no subject)
while at it, run contrib/gui. when a package is duplicated, it should be clear from the package list on the left. Suppression of duplication is incidental. What I believe is missing from sources is a simple mechanism to see if something I'm already somewhat familiar with has been ported successfully. If not, then it would be useful if there was a simple mechanism to add a successful port to the global availability, with some moderation to discourage malicious additions. As for the moderation itself, the simplest form I can think of is simply that a score be kept of successful use, reset to zero when a new version is released. Other options may prove more practical. I suspect that all the resistance I encounter is more knee-jerk reaction to a request for a little bit of additional discipline than real objection to the concept. And the usual 9fans culture of wanting code rather than discussion. Unfortunately, this is one piece of code that ought to be planned rather than hacked together: it's hard to reverse the implementation if it turns out badly. ++L
Re: [9fans] (no subject)
code-code is better than talk-talk :-) (absolutely no offense intended, I just enjoyed the phrase!) i.e. I think a proof of concept will get your further than discussion. It seems to me you could even make a copy of the sources tree, set up your idea, and put it out there for people to try. I only strongly suggest that you put the contrib gui at the heart of what the user sees. thanks ron
Re: [9fans] (no subject)
Thing is, the port hierarchy (hereto I used NetBSD package system for the same concept) provides both the hierarchical structure I believe is needed to minimise duplication and a description file to search for concepts rather than file names. So, yes, I agree with a portion of your suggestion, but not your suggested implementation. In practice, all I'm proposing is replacing directories owned by individual contributors with a hierarchy that undergoes a useful amount of validation. this is exactly the reason you need to try contrib. if you want to make an überpackage, just make your überpackage depend on everything else. it doesn't need to have any files of its own. i've verified this. it would be a trivial extension (like 2 lines) to have contrib/install recursively install dependencies. - erik
Re: [9fans] (no subject)
I only strongly suggest that you put the contrib gui at the heart of what the user sees. Agreed wholeheartedly. Thing is, It's autoconf that needs careful redesign: even though contrib/replica is not as comprehensive as a revision control even when backed by venti, it is unmatched for Plan 9 purposes. But that leaves one still at the mercy of autoconf and it seems to me that the trigger for change ought to come from Plan 9. And in case I haven't yet made it clear, I don't have a good enough picture in my head to suggest a replacement. In fact, Steve Simon and fgb both have done great work to help there, but I feel the real answer is still at large. And if anyone should find this bait irresistible, keep in mind that the final product needs to be portable across at least a few OS platforms as well as hardware architectures and that it should probably be targetting the Plan 9 toolchain rather than GCC (G++ is a problem, of course, but Go may help there as at least with the NetBSD packages, the number of C++ offerings is limited.). ++L
Re: [9fans] (no subject)
if you want to make an überpackage, just make your überpackage depend on everything else. Is this what I wanted to know all along? ++L
Re: [9fans] (no subject)
On Fri Jul 24 13:20:17 EDT 2009, quans...@quanstro.net wrote: - 9fans. sorry. i don't know what marshal did to me. - erik
Re: [9fans] (no subject)
Give or take that all the executables fail, I have enough MINGW binutils from the NetBSD package to convince me that MINGW can be built and no doubt some debugging will soon take care of the stumbling blocks. It is true that debugging is going to be hard without symbol information (one more good reason to go the ELF route for my GCC model, in my opinion) and that I need a real implementation of the SEEK syscall before I start on the debugging, but these are surmountable obstacles. I guess you need to keep holding thumbs. Sorry, this went completely over my head. It sounds like you have achieved somthing impressive but I'am afraid I don't understand. -Steve
Re: [9fans] (no subject)
1) build mingw for plan9 Give or take that all the executables fail, I have enough MINGW binutils from the NetBSD package to convince me that MINGW can be built and no doubt some debugging will soon take care of the stumbling blocks. My apologies, this was meant to be private mail to Steve Simon. ++L
Re: [9fans] (no subject)
1) build mingw for plan9 Give or take that all the executables fail, I have enough MINGW binutils from the NetBSD package to convince me that MINGW can be built and no doubt some debugging will soon take care of the stumbling blocks. It is true that debugging is going to be hard without symbol information (one more good reason to go the ELF route for my GCC model, in my opinion) and that I need a real implementation of the SEEK syscall before I start on the debugging, but these are surmountable obstacles. I guess you need to keep holding thumbs. ++L PS: Oh, and Linuxemu may have to become BSDemu in due course, too. But I have too much to learn before I can treat that as an option.
Re: [9fans] (no subject)
lcc is nothing like as hard to compile as gcc (which has got worse, much worse, over the years). funnily enough, my gcc bootstrap compilation is still going (on a multi-core linux machine). it started over an hour ago. bizarre.
Re: [9fans] (no subject)
2009/4/23 Charles Forsyth fors...@terzarima.net: lcc is nothing like as hard to compile as gcc (which has got worse, much worse, over the years). funnily enough, my gcc bootstrap compilation is still going (on a multi-core linux machine). it started over an hour ago. bizarre. (my apologies for diverging the topic from both Plan 9 and the original question) What languages are you compiling? on my 3-core amd processor, it takes 10 minutes to compile c and c++. Steve: I know there are some really good resources about PE online, as well as several on MSDN that might contain more information about what Windows will actually do with a PE. I don't believe they are required to be relocatable, as even Windows 7 will still run Win9x binaries, and I'm fairly sure that there wasn't the ability to relocate executables back then. --dho
Re: [9fans] (no subject)
lcc is nothing like as hard to compile as gcc (which has got worse, much worse, over the years). funnily enough, my gcc bootstrap compilation is still going (on a multi-core linux machine). it started over an hour ago. bizarre. ... and when I eventually completed it under NetBSD, a few months back, the GAS pass failed with a segmentation violation. I wonder if it's been fixed since... ++L
Re: [9fans] (no subject)
back, the GAS pass failed with a segmentation violation. unfortunate! that's gotta hurt. ☺ - erik
Re: [9fans] (no subject)
On Thu, Apr 23, 2009 at 10:17 AM, erik quanstrom quans...@quanstro.netwrote: back, the GAS pass failed with a segmentation violation. unfortunate! that's gotta hurt. ☺ - erik Seems like you'd have to change your pants after a GAS pass segmentation violation.
Re: [9fans] (no subject)
anyone know of any other (simpler) options? Inferno utils? The binaries for NT? ++L
Re: [9fans] (no subject)
On Mon, Feb 23, 2009 at 4:29 PM, mattmob...@proweb.co.uk wrote: oops I forgot carriage returns get converted in email imagine # is a ^m pu% /n/sources/contrib/maht/rc/til abc¯def#123# emit abc def 123 : emit4 emit emit emit emit # 1 2 3 emit4 321 1 # swp 3 2 swp dup dup drop emit4 emit 3321 does this make sense / of interest to anyone :) sure but what's the point? Couldn't you grab one of the forth interpreters? ron
Re: [9fans] (no subject)
On Feb 23, 2009, at 6:48 PM, ron minnich rminn...@gmail.com wrote: sure but what's the point? Couldn't you grab one of the forth interpreters? Forth on Plan 9? That would be great. Then I'd have a bit more leverage in getting a new PIC based board to talk with Plan 9. (brucee's recent 9p example's notwithstanding) -jas
Re: [9fans] (no subject)
Jeff Sickel wrote: On Feb 23, 2009, at 6:48 PM, ron minnich rminn...@gmail.com wrote: sure but what's the point? Couldn't you grab one of the forth interpreters? Forth on Plan 9? That would be great. Then I'd have a bit more leverage in getting a new PIC based board to talk with Plan 9. (brucee's recent 9p example's notwithstanding) -jas Someone in freenode #forth said they'd just ported 4tH to Plan 9. I don't log IRC, wish I'd payed more attention. Maybe they've shared it by now. I don't doubt them, as 4tH's source comes with instructions on how to build with no libs and only 2 syscalls (1 is 'printf' replacement). sixforty
Re: [9fans] (no subject)
fgb has ported 4th; it's in his contrib dir, both as a tar file and a contrib package. i thought i remembered seeing another, but it's not on the contrib index page.
Re: [9fans] (no subject) [troff / man bug]
Suject: scuzz(8) man bug man -P, man or the magicman html versions all have a similar bug. the .IR macro appends a a spurious .}f. SEE ALSO sd(3) Small Computer System Interface - 2 (X3T9.2/86-109), .}f shortening the italic portion by 4 characters makes the problem go away. does anybody know what's wrong here? - erik
Re: [9fans] (no subject)
On Dec 6, 2008, at 11:28 AM, Dave Eckhardt wrote: More globally, if the high adoption rate of NFS is an argument in favor of its architecture, It is most definitely not. At least in my opinion. However, adoption is the only thing that I know of that can potentially justify excessive *engineering* complexity. Personally I'd call any system that is overly complex *and* has a low adoption rate (for whatever reason) after at least 5 years of having been exposed to the market place, a theoretical artifact of computer science. The usefulness of which is commonly measured by the number of good CS papers dedicated to studying it. I do understand that for you AFS also has a huge redeeming factor of being useful in solving a particular problem. It is hard to say whether it is the best tool for the job, or the only one (it would be interesting if you answered to Erik's question on this list). But perhaps the difference of opinion here stems from the fact that to *you* it does feel like an enterprise grade to me (as an outside observer) the enterprise grade must be vetted by the market place. Hence my interest in the adoption rate. and the low adoption rate of AFS is an argument against its architecture, why are you reading a Plan 9 mailing list...? This one is easy: Plan 9 (and 9P in particular) doesn't have to have the redeeming quality of high adoption rate in order to justify an excessive engineering complexity. It is not complex at all. It is small and elegant. Whether that compactness and elegance sometimes prevents it from being considered enterprise grade is an open question (at least for me it is). The experience of Coraid suggests that it might actually be a nice tool even for those kinds of problems. To some extent, the popularity of NFS (is there any NAS box that talks AFS?) and Linux is one big testament to the power of good enough or worse is better. Designing enterprise grade things is very hard work. Implementing them is even harder. The good news is that it pays well. The bad news is that you have to be really brave to withstand the fear of being obsolete by changing requirements. I don't get this. I don't follow the NFS protocol development carefully, Just to clarify: this thread is really not about me defending NFS. I can't call it elegant. But I can certainly call it good enough. So at least it succeeds as a lowest common denominator for data sharing. Well, so far at least. That is, I think the requirements are *not* changing, but rather that NFS is slowly realizing that those things *are* requirements. Agreed. And I believe both FSs do miss the point. NFS4 especially so. The question is really not about the most efficient implementation of POSIX filesystem semantics over the network, but rather whether POSIX expectations are reasonable in the first place. Plan 9 was bold enough to simply discount some of those expectations and the resulting system proved to be much better than what a typical UNIX provides these days. Now, what would be really interesting is to see is how the kinds of requirements that pNFS has can be satisfied with Plan9/9P approach. Thanks, Roman.
Re: [9fans] (no subject)
This one is easy: Plan 9 (and 9P in particular) doesn't have to have the redeeming quality of high adoption rate in order to justify an excessive engineering complexity. It is not complex at all. It is small and elegant. Whether that compactness and elegance sometimes prevents it from being considered enterprise grade is an open question (at least for me it is). The experience of Coraid suggests that it might actually be a nice tool even for those kinds of problems. i believe the term of art is enterpriseiness. - erik
Re: [9fans] (no subject)
Hello, If such an attack continues for some minutes and the server does not reject the connections the server will create thousands of smtpd processes and might be hung up. Kenji Arisawa On 2008/11/22, at 3:28, erik quanstrom wrote: Subjet: email attacks since our friends in sweeden helped out our spammer friends get back on line, i've seen a lot more attacks. today i've been getting ~10 connections/sec. fortunately its from a small number of machines, so this trick helps alot: /n/dump/2008/1121/sys/src/cmd/upas/smtp/smtpd.c:348,353 - smtpd.c: 348,355 if(!qflag) syslog(0, smtpd, Hung up on %s; claimed to be %s, nci-rsys, him); + if(Dflag) + sleep(delaysecs()*1000); reply(554 5.7.0 Liar!\r\n); exits(client pretended to be us); return; oddly, i've found that adding a few of the hosts as -k flags stops the attack entirely. - erik
Re: [9fans] (no subject)
Hello, If such an attack continues for some minutes and the server does not reject the connections the server will create thousands of smtpd processes and might be hung up. Kenji Arisawa but that's not what happens. waiting for 15 seconds decreases the load to near zero. it's not a ddos attack. in fact, most of the trouble came from a single zombie. if it were, i would need to see more than 10 connections/second before i would have a significant backlog. i can easily afford 150 smtpds sleeping. - erik