On Thursday, January 27, 2022 7:17:10 PM EST Steffen Möller wrote: > On 28.01.22 00:31, gene heskett wrote: > > On Thursday, January 27, 2022 2:22:43 PM EST Steffen Möller wrote: > >> On 27.01.22 20:03, gene heskett wrote: > >>> On Thursday, January 27, 2022 9:54:54 AM EST Steffen Möller wrote: > >>>> Fresh start! > >>>> > >>>> Dear Gene, > >>>> > >>>> I would like to catch the problem before you start building and > >>>> also > >>>> exclude the possibility that somehow the code base of yours is > >>>> affected by your previous checkout - just because I cannot inspect > >>>> your machine from here. Once that was successful, yes, then this > >>>> can be optimized. > >>>> > >>>> First thing is that the system needs to be truly updated, nothing > >>>> half-ish. > >>> > >>> Remember Steffen, that this sd card was A, new, and b, written with > >>> dd > >>> using 2021-10-30-raspios-bullseye-armhf-full.img, > >>> > >>> sd card then put in the pi and booted, after I had fixed the no > >>> network problem by filling in the defaults for a static network by > >>> putting my hosts file over the default, editing /etc/hostname to be > >>> the same as the buster install it would replace, and filling in and > >>> uncommenting the bottom of its /etc/dhcpcd.conf file to match the > >>> buster net config. It was then booted, I assigned the country, > >>> keyboard and other first boot things in raspi-config. rebooted, at > >>> which point it ran the update/ upgrade stuff bringing it up to date > >>> by upgrading 129 pkgs then. The apt update/apt upgrade -y has been > >>> done several more times, and just now replaced 5 python pkgs. > >>> > >>> sudo apt dist-upgrade just returned: > >>> ============================= > >>> pi@rpi4:/media/pi/workspace $ sudo apt dist-upgrade -y > >>> Reading package lists... Done > >>> Building dependency tree... Done > >>> Reading state information... Done > >>> Calculating upgrade... Done > >>> The following packages were automatically installed and are no > >>> longer > >>> > >>> required: > >>> dctrl-tools dkms libfuse2 libxtables-dev > >>> raspberrypi-kernel-headers > >>> > >>> Use 'sudo apt autoremove' to remove them. > >>> 0 upgraded, 0 newly installed, 0 to remove and 0 not upgraded. > >>> ============================= > >>> So it is, ANAICT, a fully uptodate bullseye install. > >>> > >>> In the middle of that, I mounted the drive I had built a > >>> 5.16.2-rt19 > >>> kernel on and installed that, so a uname -a now returns: > >>> =========================== > >>> Linux rpi4 5.16.2-rt19-rt19-v7l+ #3 SMP PREEMPT_RT Tue Jan 25 > >>> 01:14:16 > >>> EST 2022 armv7l GNU/Linux > >>> =========================== > >>> Almost as bleeding edge as has been announced on the linux-rt list. > >>> I have since ran the first of my scripts to build master, then > >>> scanned > >>> the output for missing dependency's, installing those I spot in the > >>> back trace. > >>> > >>> I have not adjusted anything in the config, or debian directories > >>> of > >>> this git clone which is of the raspberry/linux. I have been doing > >>> this build since jessie days so I'm not new to this. At one point > >>> pi > >>> stuff was being built on an odroid C2 at the buildbot, but crashed > >>> several times a week and has been replaced with an rpi4 I believe, > >>> same as this one. > >>> > >>> I faintly recall having to do something for buster but at 87 yo I > >>> do > >>> not recall what it was I had to do back then. And I have not noted > >>> the build your own recipe in our wiki as having been updated since > >>> wheezy, so it is, shall we say, a bit long in the tooth in 2022. :) > >>> > >>> And the next missing dependency is "convert", and its a showstopper > >>> for configure. > >> > >> I was not aware of "convert". You have done everything just fine. > >> > >>>> Please do > >>>> > >>>> sudo apt update > >>> > >>> that showed 5 pkgs could be upgraded which I did. > >>> > >>>> sudo apt -u dist-upgrade > >>> > >>> see above > >>> > >>>> Anything surprising/weird/many packages listed? Then please tell > >>>> me > >>>> or > >>>> continue with "yes". > >>>> > >>>> Wherever you have the disk space please then do a > >>> > >>> I have the space, its a 240gig SSD > >>> > >>>> git clone https://github.com/LinuxCNC/linuxcnc.git > >>>> bullseye-linuxcnc > >>> > >>> Will take an hour or more, my net connection is leisurely. > >>> > >>>> cd new-dir-name > >>>> > >>>> python3 --version # should be more recent than 3.7 > >>> > >>> pi@rpi4:/media/pi/workspace $ python3 --version > >>> Python 3.9.2 > >>> > >>>> Please ping me again once you got to this stage. > >>> > >>> And configure still bails out: > >>> checking for convert... none > >>> > >>> configure: error: no convert, documentation cannot be built > >> > >> You are two steps ahead :) I presume you ran > >> > >> debian/configure > >> > >> but just do it again, please, so I know what was done. The "uspace" > >> argument is the default, so just run as shown above. > >> > >> We now have the debian/control file. This debian/control file > >> declares > >> the packages that are required to run the package, but especially > >> also > >> the packages that are required to build the package. > >> > >> Now run > >> > >> dpkg-buildpackage > >> > >> and it will check that debian/control file to see if all packages > >> are > >> indeed installed that need to be installed to build the packages. > >> > >> When you invoke this now then it will fail (so I hope) because the > >> package "imagemagick" is missing which then also provides > >> /usr/bin/convert. There are likely other packages, too, that > >> configure > >> would identify as missing if it was not already halting after > >> checking > >> for "convert". So, dpkg-buildpacakge will list all the packages that > >> are missing. Please install those and then run dpkg-buildpackage > >> again. > >> > >> LinuxCNC has a bit of a problem with interrupted builds. There is a > >> chance that dpkg-buildpackage fails because files existing that are > >> not in the original tarball. But we will get to that - now it is me > >> who is ahead of himself :) > >> > >> Best, > >> > >> Steffen > > > > I installed imagemagick and all but one mesa util apt couldn't find, > > did a make clean > > dpkg-buildpackage invokes a cleaning itself. This should not be > required. In theory. > > > and reran configure, > > Which configure? The ./configure ? Maybe this works. Maybe not. > Configure takes options and for proper debian packages these need to be > set properly :) It is not magic, how it is done you can inspect in the > file debian/rules, which is derived from debian/rules.in by a call to > debian/configure. > > Please do a > > git reset --hard > > which should bring you back to pristine master branch. Please also do a > > git status > > and remove any files that are listed as not tracked (there should not > be any, just to be sure). > > Then start over with debian/configure (the debian one, not the one in > your working directory, which should now no longer exist since it was > generated). No need to directly invoke any configure or any make from > here on. The file debian/rules is a Makefile itself and knows how to > invoke the real Makefile. And that debian/rules file is invoked via > dpkg-buildpackage. > > So, after debian/configure please run dpkg-buildpackage again and this > time it should complete. If not, well, then something is special on > your RPi platform. We'll get there, eventually. > > Best, > Steffen > > > which didn't fuss about anything, then > > fired off dpkg-buildpackage again which spat out its usual warnings > > from strncopy for about 40 minutes before bailing out with this > > lengthy wall of fuss: > > ===================================== > > /usr/bin/ld: objects/emc/task/taskclass.o: in function > > `boost::python::converter::arg_to_python<char [...] And despite installing some more "boost python3" stuff, I'm still stuck looking at a wall about 40 feet high, complaining about undefined references in boost::python::*::*::* where you could use the whole dictionary for the *, like: /usr/bin/ld: ../lib/librs274.so.0: undefined reference to `boost::python::detail::str_base::str_base(char const*)'
Basically if its boost::python, its an undefined reference. Since the syntax shows a perl like format, I've even search for boost stuff written in perl w/o stumbling over anything that looks like a likely suspect. Could it be as simple as 2to3 not being installed? It wasn't, tis now. Didn't help. Same wall of undefined references. ending in: /usr/bin/ld: ../lib/librs274.so.0: undefined reference to `boost::python::detail::tuple_base::tuple_base(boost::python::api::object const&)' collect2: error: ld returned 1 exit status make[2]: *** [emc/task/Submakefile:37: ../bin/milltask] Error 1 make[2]: Leaving directory '/media/pi/workspace/bullseye-linuxcnc/src' make[1]: *** [debian/rules:51: override_dh_auto_build-arch] Error 2 make[1]: Leaving directory '/media/pi/workspace/bullseye-linuxcnc' make: *** [debian/rules:39: build] Error 2 dpkg-buildpackage: error: debian/rules build subprocess returned exit status 2 Your turn :) I am at your command. Thanks Steffen. Take care and 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/gene> _______________________________________________ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
