#linuxcnc-devel Logs

May 05 2020

#linuxcnc-devel Calendar

02:39 AM linuxcnc-build: build #4932 of 1400.rip-wheezy-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1400.rip-wheezy-i386/builds/4932 blamelist: Phillip Carter <phillcarter54@gmail.com>
03:56 AM linuxcnc-build: build #6793 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/6793 blamelist: Phillip Carter <phillcarter54@gmail.com>
11:03 AM seb_kuzminsky: linuxcnc-build: force build --branch=2.8 1650.rip-buster-rtpreempt-rpi4
11:03 AM linuxcnc-build: build #156 forced
11:03 AM linuxcnc-build: I'll give a shout when the build finishes
11:36 AM linuxcnc-build: Hey! build 1650.rip-buster-rtpreempt-rpi4 #156 is complete: Warnings [8warnings compile]
11:36 AM linuxcnc-build: Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1650.rip-buster-rtpreempt-rpi4/builds/156
12:38 PM cradek: github emails are now all full of bogus html quoting in the text/plain part
01:32 PM mozmck: Well, they are now owned by microsoft too :-)
01:59 PM rene_dev_: seb_kuzminsky jepler what do you think about changing at least one build to python3?
02:07 PM jepler: we could do it in buildbot or in github actions. either way, it looks like there's more work to do (sigh, of course there is). github actions --and, I think? buildbot -- depends on some scripts to install the right dependencies. For github actions, the So next step would be to extend debian/configure so that it has a way to create a debian source package that only has python3 dependencies. Then you extend
02:07 PM jepler: scripts/travis-install-build-deps.sh so it can be invoked in the "python3 way" if requested; scripts/travis-build-test needs similar treatment. Finally, you can add a new stanza in .github/workflows/ci.yml under "jobs:", perhaps called "main3:" which puts it all together
02:08 PM jepler: the good thing is that you can push this to your personal github fork and see the results within its actions tab, and verify everything before creating a pull request
02:13 PM jepler: and say you want to just completely ream any of those things and start with fresh python3 versions that jettison baggage, that's fine too -- debian/configure3 or whatever. but since the true deliverable of linuxcnc is installable debian packages, keep that in mind as the goal.
02:19 PM jepler: scripts that appear to still refer to python2: https://gist.github.com/jepler/967e153b9badc80b9f101d82423c5a35
02:19 PM jepler: in #!-lines
02:22 PM jepler: https://gist.github.com/88c25bf5d4211bfdb86e5c58de5ab18e straw conversion of debian/configure to add python3 support on debian 10
02:35 PM rene_dev_: jepler is the linuxcnc-python script available on a packaged and installed linuxcnc?
02:39 PM jepler: clearly it'll need to be?
02:39 PM jepler: well, that depends; if only tests use it then it wouldn't need to be
02:40 PM jepler: I don't recall whether the patch I sent you would have caused it to be installed, I'd have to build or download a package and check
03:01 PM seb_kuzminsky: i want the buildslaves to be configured as close as possible to how a user would have their machine configured, so i'm reluctant to change /etc/alternatives or uninstall default packages or other changes like that
03:01 PM seb_kuzminsky: however i also really appreciate the work you've done on python3 support, and i want to help that effort
03:02 PM seb_kuzminsky: does bullseye use py3 by default? if so i could add a buildslave with that
03:02 PM jepler: I don't know about bullseye
03:02 PM seb_kuzminsky: if not, another option i guess would be to add a buildslave specifically for testing python3, and modify the install so its /usr/bin/python points to python3, even though that's not the way most of our users would experience it
03:03 PM jepler: For now it may be that putting this in github actions is the way to go, not buildbot.
03:03 PM jepler: especially since github actions has more transparency for developers
03:03 PM jepler: maybe that's not the word I mean, but basically they can see the checks without pushing to the linuxcnc fork itself
03:03 PM seb_kuzminsky: yeah, that is a nice feature of github
03:05 PM jepler: but we do also want to know that the python3 packages build at all ... whether that would be best done on github or on buildbot I don't know
03:05 PM jepler: bullseye's `python` package is still python2.7.
03:05 PM jepler: https://packages.debian.org/source/bullseye/python-defaults
03:05 PM seb_kuzminsky: yep
03:06 PM seb_kuzminsky: no free lunch for us
03:13 PM seb_kuzminsky: back in 2019 the goal was to release bullseye with python3 only: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=931659
03:15 PM seb_kuzminsky: or maybe at least with python3 as default, and python2 available if you really wanted it?
03:16 PM seb_kuzminsky: looks like there's still lots of work before py2 can be removed: https://release.debian.org/transitions/html/python2-rm.html
03:37 PM -!- #linuxcnc-devel mode set to +v by ChanServ
03:38 PM seb_kuzminsky: i chatted with the folks on #debian-python and they had this advise:
03:38 PM seb_kuzminsky: don't wait for /usr/bin/python to become python3, that may never happen, or at least not for a long time
03:38 PM jepler: https://release.debian.org/transitions/html/python2-rm.html
03:39 PM seb_kuzminsky: instead, explicitly shebang /usr/bin/python3 when you're ready, that will always give you what you expect
03:39 PM jepler: we might decide as a project that master will be python3-only and then we would change our configure calls when building packages .. but that might be too much to ask yet, with python3 support so new to us
03:39 PM jepler: whoops not sure how I pasted that URL again
03:40 PM seb_kuzminsky: jepler: i think that's the correct way to do it - we "manually" and explicitly switch some branch to python3 when we feel like it
03:40 PM seb_kuzminsky: i guess when that Minimum Software Versions wiki page tells us we no longer care about platforms that are so old that they don't have a functional python3
03:41 PM cerna: Expecting /usr/bin/python to be specific version may also bite people who build for different distros. (There was some discussion about it recently.) I know that Alpine already tried it/was playing with it.
03:42 PM cerna: Changing /usr/bin/python to python 3 default.
03:42 PM seb_kuzminsky: rene_dev_: do you know what the minimum python3 version for your stuff is?
03:43 PM seb_kuzminsky: wheezy has 3.2, jessie has 3.4, stretch has 3.5, and buster has 3.7
03:44 PM jepler: cerna: your concern about this is noted, thanks. In fact, we switched to having the Debian package configure --with-python=python2.7 just recently for this reason.
03:44 PM jepler: at least I think that got merged
03:46 PM jepler: (and why we introduced the linuxcnc-python wrapper script, etc)
03:47 PM jepler: however, some other-distro/alternate universe advice to use the program name "python2" doesn't work for us, which is why we went with python2.7 in our debian configure invocations. Anyone can invoke configure with the python executable name they prefer and that should get carried through to everything that is 'touched' by the build process, or which uses the linuxcnc-python wrapper
03:47 PM jepler: however we still have 180+ scripts that invoke a non-numbered or explicitly-python2 version in a #!-line, gist pasted earlier
03:51 PM cerna: jepler: OK, just a note. I have no (strong) feeling one way or another.
03:52 PM cerna: About Github Actions, I think it can work great for the RT_Preempt/Vanilla linux developer testing.
03:53 PM cerna: Given that developer mostly would want to test amd64 only before creating a pull request.
04:00 PM rene_dev_: seb_kuzminsky no, that is to be figured out. andy did figure out that wheezy si too old.
04:02 PM rene_dev_: jepler I only added it to the tests, no other programs.
04:03 PM rene_dev_: seb_kuzminsky jepler I wanted to be in a way that it does not depend on wether python points to 2 or 3, as this depends on the distro.
04:18 PM seb_kuzminsky: we build some python modules from C source code (hal.so and linuxcnc.so, possibly others)
04:18 PM seb_kuzminsky: i guess those have to be built for either python2 or python3, is that controlled by the src/configure script via --with-python=XXX?
04:20 PM rene_dev_: yes, they will only work with the version they were compiled for.
04:23 PM jepler: At this point -- after the EOL of python2 --- trying to support 2 and 3 from a single build is NOT worth spending time over.
04:24 PM jepler: so I am fully behind rene_dev_ 's approach there
04:24 PM seb_kuzminsky: agreed
04:25 PM andypugh: Every OS that we support LinuxCNC on has some version of Py3, so I would suggest that we make Python depend on 3 and ignore 2
04:25 PM seb_kuzminsky: old branches, 2.7 & 2.8, will be python2 only; future branches will be python3 only
04:25 PM seb_kuzminsky: andypugh: the problem is apparently wheezy's python 3.2 is too old
04:25 PM andypugh: But I am not fully clued-up about the technicalities of that
04:25 PM seb_kuzminsky: we'll have to drop wheezy when we switch to python3
04:25 PM andypugh: Wheezy is too old.
04:26 PM seb_kuzminsky: and pour out a bottle of way oil for our dead homies
04:27 PM andypugh: If we can get RTAI working then there is no reason to stick with Wheey, 3 years out of LTS
04:27 PM andypugh: Though I do wonder if we can packport a newer Python to it?
04:27 PM seb_kuzminsky: the only reason would be if we wanted to make easy upgrades for folks with existing machines
04:28 PM andypugh: There is a lot of goodness in 2.8 and Master.
04:28 PM andypugh: But not so much (at the moment) between 2.8 and Master
04:28 PM seb_kuzminsky: backporting a newer python3 to wheezy is an interesting idea, but i fear it's a serious can of worms
04:29 PM seb_kuzminsky: you're not planning to drop wheezy in 2.8 are you? just for master?
04:29 PM andypugh: No, 2.8 is still Py2 and Wheezy
04:29 PM seb_kuzminsky: ok good
04:29 PM jepler: If you have a support contract, Wheezy is still ETLS for 2 more months https://wiki.debian.org/LTS/Extended
04:30 PM jepler: (not defending it as something we should continue to target)
04:30 PM seb_kuzminsky: so we can maintain 2.7 & 2.8 with py2 and wheezy until we get sick of it, and we can jump to stretch and py3.5 for master
04:30 PM seb_kuzminsky: i assume python3.5 is new enough? rene_dev_ ?
04:32 PM andypugh: I have a wheezy VM here. Let me see how it fares with Py3.4
04:33 PM jepler: The supported version promises we made for 2.8 made sense when we imagined it would come out in 2016 or 2017. It's great if we can meet them or come close to meeting them, even though they're increasingly absurd. But, except for avoiding crossing off RTAI entirely, there's no reason not to make a less extreme set of promises for 2.9 / master
04:33 PM andypugh: We are working on RTAI for Buster / Bulseye
04:34 PM andypugh: It’s just a tiny bit crashy when you unload it. Seems solid enough actually running LinuxCNC
04:35 PM linuxcnc-build: build #6773 of 1300.rip-precise-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1300.rip-precise-i386/builds/6773 blamelist: Jeff Epler <jepler@gmail.com>
04:36 PM linuxcnc-build: build #6776 of 1306.rip-precise-amd64 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1306.rip-precise-amd64/builds/6776 blamelist: Jeff Epler <jepler@gmail.com>
04:36 PM jepler: failure looks real, must be due to whatever I just pushed on behalf of pcw
04:37 PM jepler: hostmot2: dlopen: /home/jepler/src/linuxcnc/rtlib/hostmot2.so: undefined symbol: hm2_rcpwmgen_prepare_tram_write
04:37 PM jepler: hmmmph
04:39 PM pcw_home: oops sorry...
04:39 PM jepler: my fault, carry on
04:39 PM jepler: bad testing hygene on my part
04:40 PM andypugh: seb_kuzminsky: Several days ago you were experimenting with default boot kernels on multi-boot PCs. Did you manage to make it work?
04:41 PM linuxcnc-build: build #5994 of 1301.rip-precise-rtai-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1301.rip-precise-rtai-i386/builds/5994 blamelist: Jeff Epler <jepler@gmail.com>
04:41 PM andypugh: As I am involved in crash-testing, and the actual machine is upstairs, it would save me a great deal of tedium if I could figure it out.
04:41 PM seb_kuzminsky: andypugh: yeah, it's been working well for me
04:42 PM linuxcnc-build: build #4935 of 1403.rip-wheezy-amd64 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1403.rip-wheezy-amd64/builds/4935 blamelist: Jeff Epler <jepler@gmail.com>
04:42 PM linuxcnc-build: build #4934 of 1400.rip-wheezy-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1400.rip-wheezy-i386/builds/4934 blamelist: Jeff Epler <jepler@gmail.com>
04:42 PM andypugh: Did you choose by number or by name?
04:43 PM seb_kuzminsky: by number
04:43 PM seb_kuzminsky: hang on, i'll look up the details in my logs
04:43 PM seb_kuzminsky: andypugh: my /etc/default/grub says:
04:43 PM seb_kuzminsky: GRUB_DEFAULT="1>4"
04:43 PM seb_kuzminsky: that machine boots automatically to the kernel i want
04:43 PM seb_kuzminsky: that's the 5th kernel from the submenu that is item 2 in the main menu
04:43 PM seb_kuzminsky: this is my main menu: http://highlab.com/~seb/linuxcnc/grub-main-menu.jpg
04:43 PM seb_kuzminsky: this is the submenu: http://highlab.com/~seb/linuxcnc/grub-submenu.jpg
04:43 PM seb_kuzminsky: and this is my GRUB_DEFAULT: http://highlab.com/~seb/linuxcnc/grub-default.jpg
04:44 PM seb_kuzminsky: the double-quotes around the "1>4" are important
04:44 PM seb_kuzminsky: hope that helps
04:45 PM andypugh: Possibly. Though in my case I would be wanting the 4th (index 3) of the first menu.
04:48 PM linuxcnc-build: build #4454 of 1402.rip-wheezy-rtpreempt-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1402.rip-wheezy-rtpreempt-i386/builds/4454 blamelist: Jeff Epler <jepler@gmail.com>
04:49 PM linuxcnc-build: build #5138 of 1404.rip-wheezy-rtpreempt-amd64 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1404.rip-wheezy-rtpreempt-amd64/builds/5138 blamelist: Jeff Epler <jepler@gmail.com>
04:54 PM linuxcnc-build: build #158 of 1650.rip-buster-rtpreempt-rpi4 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1650.rip-buster-rtpreempt-rpi4/builds/158 blamelist: Jeff Epler <jepler@gmail.com>
05:01 PM seb_kuzminsky: so back to the python3 thing, it sounds like we the next step is to add some commits to rene_dev_ 's branch to 1) disable wheezy and older in scripts/platform-is-supported, and 2) switch src/configure to default to python3
05:02 PM seb_kuzminsky: then push that branch to the LinuxCNC github so the buildbot tries it out on all the still-supported platforms and we'll see what breaks next
05:02 PM seb_kuzminsky: does that capture it?
05:03 PM jepler: rene's stuff is merged to master branch
05:03 PM seb_kuzminsky: oh!
05:03 PM rene_dev_: I spend some time to make everything 2 and 3 compatible
05:03 PM seb_kuzminsky: that shows how out of the loop i am
05:04 PM jepler: the question is, how do we make sure that the new python3 stuff doesn't bitrot, by feeding it to CI in some capacity
05:04 PM rene_dev_: takes suprisingly little #ifdefs, most of that is handled by py3c
05:04 PM seb_kuzminsky: is it too early to make the two changes i suggested to master?
05:05 PM jepler: well, 75% of all the UIs don't work anymore once you switch to python3
05:05 PM seb_kuzminsky: maybe src/configure could default to python3 if 3.5 or newer is available (or whatever the magic number is), and default to python2 if it's not?
05:05 PM jepler: not because of python3 per se, but because they have dependencies that don't exist in python3, such as gtk2, if I understand correctly
05:05 PM rene_dev_: gtk UIs cant worj with python3 :(
05:05 PM seb_kuzminsky: ooookay, the dim lightbulb is slowly starting to flicker to life in my head
05:06 PM jepler: (so I think axis and keystick are what "work" with python3; I haven't tried it myself)
05:06 PM jepler: I'm about to split, sorry.
05:06 PM rene_dev_: axis, tcl things
05:06 PM rene_dev_: qt UIs should be easy to get working
05:08 PM rene_dev_: gtk works with gobject introspection, but is unsupported. the real problem is gtkgl, that isnt available.
05:10 PM rene_dev_: the successor, gtkglarea, is only available in gtk3. I have a branch where I ported everything to gtk3, but gtkglarea only creates a core opengl profile, so the fixed function pipeline in linuxcnc doesnt work.
05:11 PM rene_dev_: you cant force it to create a legacy profile.
05:11 PM seb_kuzminsky: it's easy enough to pass custom arguments to debian/configure and src/configure in the buildbot, on a per-builder basis
05:12 PM seb_kuzminsky: so i could add a new buildslave and make it build rip with the right --with-python=python3 argument to src/configure
05:13 PM seb_kuzminsky: it won't build packages for now, but it's something at least
05:13 PM seb_kuzminsky: would that do what you need, rene_dev_ ?
05:14 PM rene_dev_: there isnt any way to create a legacy context in gtk3, and even if there would be, it would stop working with gtk4 which will be released later this year, as it uses opengl to render everything.
05:15 PM rene_dev_: seb_kuzminsky yes :) it also needs the bootst libs set, as it cant find them automatically
05:15 PM rene_dev_: and run the tests, to make sure the python code keeps working.
05:18 PM jepler: it's too bad we collected so much technical debt, python2 + gtk2 + opengl1.2
05:18 PM jepler: (not to mention tk and nml, but one we build ourselves and the other doesn't have the strength left for a cataclysmic change like Python did)
05:19 PM rene_dev_: python is solved, and gtk almost :D
05:19 PM rene_dev_: gmocappy runs with gtk3 and python3, apart form the gremlin
05:20 PM rene_dev_: gscreen and touchy require some work
05:20 PM rene_dev_: that also solves the glade problem
05:22 PM jepler: I should work on the gl stuff
05:24 PM rene_dev_: Im having a gaming guy workig on it :)
05:24 PM rene_dev_: opengl 3.3 is probably a good target.
05:25 PM rene_dev_: jepler have you used anything beyond fixed function pipeline?
05:25 PM jepler: Oh great
05:25 PM jepler: Yes but not well
05:26 PM rene_dev_: you can checkout my branch, and start gmoccapy. the gremlin is using a gtkglarea that does nothing.
05:26 PM rene_dev_: https://github.com/rene-dev/linuxcnc/tree/rene-gtk3
05:26 PM jepler: At my old day job we built basically an emulation layer between, and I learned enough to do they. But it's in no way like good modern openGL
05:26 PM jepler: And I don't have my own coffee to crib from now (probably for the best)
05:27 PM jepler: Ha own code
05:28 PM rene_dev_: jepler and gtkglarea has a bug, where it doesnt work with software renderers: https://gitlab.gnome.org/GNOME/gtk/issues/64
05:29 PM rene_dev_: but I already created a PR to fix that.
05:29 PM rene_dev_: I discovered that years ago when working with hazzy-m on gtk3 things
05:32 PM rene_dev_: hazzy-m TurBoss have you tried qt5 with python3?
05:34 PM TurBoss: i did ran 2 to 3 it worked mostly well
05:35 PM TurBoss: some stuff is missing lyke bytes strings changes
05:35 PM TurBoss: * some stuff is missing like bytes strings changes
05:37 PM rene_dev_: they are usually an easy fix.
05:37 PM rene_dev_: but can you get it to start?
05:38 PM linuxcnc-build: build #3401 of 1500.rip-jessie-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1500.rip-jessie-i386/builds/3401 blamelist: Jeff Epler <jepler@gmail.com>
05:38 PM TurBoss: it boots and homes
05:38 PM linuxcnc-build: build #1791 of 1610.rip-stretch-rtpreempt-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1610.rip-stretch-rtpreempt-i386/builds/1791 blamelist: Jeff Epler <jepler@gmail.com>
05:38 PM rene_dev_: cool! does that make the code 2 and 3 compatible?
05:38 PM * TurBoss uploaded an image: image.png (47KB) < https://matrix.org/_matrix/media/r0/download/jauriarts.org/shcdfHczvdlWoTqmOncFXRzI >
05:38 PM rene_dev_: if yes, can you push the changes?
05:39 PM TurBoss: thoose notset are not working stuff related to rules
05:39 PM TurBoss: they are on a branch named 2to3
05:40 PM rene_dev_: which UI is that?
05:40 PM TurBoss: mini
05:40 PM TurBoss: its included
05:40 PM linuxcnc-build: build #3401 of 1510.rip-jessie-rtpreempt-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1510.rip-jessie-rtpreempt-i386/builds/3401 blamelist: Jeff Epler <jepler@gmail.com>
05:40 PM linuxcnc-build: build #3403 of 1520.rip-jessie-amd64 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1520.rip-jessie-amd64/builds/3403 blamelist: Jeff Epler <jepler@gmail.com>
05:41 PM TurBoss: i didnte try pythno2 after that
05:41 PM TurBoss: * i didnt try pythno2 after that
05:42 PM andypugh: seb_kuzminsky: I think my problem is that I might not be updating the right grub.cfg. I booted in to Mint and the contemts of default/grub were different
05:44 PM linuxcnc-build: build #397 of 1640.rip-buster-rtpreempt-amd64 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1640.rip-buster-rtpreempt-amd64/builds/397 blamelist: Jeff Epler <jepler@gmail.com>
05:44 PM Tom_itx is now known as Tom_L
05:47 PM linuxcnc-build: build #1791 of 1630.rip-stretch-rtpreempt-amd64 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1630.rip-stretch-rtpreempt-amd64/builds/1791 blamelist: Jeff Epler <jepler@gmail.com>
05:47 PM linuxcnc-build: build #4607 of 1401.rip-wheezy-rtai-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1401.rip-wheezy-rtai-i386/builds/4607 blamelist: Jeff Epler <jepler@gmail.com>
05:49 PM linuxcnc-build: build #3404 of 1530.rip-jessie-rtpreempt-amd64 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1530.rip-jessie-rtpreempt-amd64/builds/3404 blamelist: Jeff Epler <jepler@gmail.com>
05:49 PM linuxcnc-build: build #203 of 1635.rip-buster-rtpreempt-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1635.rip-buster-rtpreempt-i386/builds/203 blamelist: Jeff Epler <jepler@gmail.com>
05:49 PM linuxcnc-build: build #6795 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/6795 blamelist: Jeff Epler <jepler@gmail.com>
05:58 PM rene_dev_: jepler it requires pyglm https://pypi.org/project/PyGLM/
05:58 PM rene_dev_: which on debian, you can only have via pip. but I recall we cant depend on packages installed via pip?
06:03 PM jepler: Nope
06:05 PM rene_dev_: so how can we use pyglm
06:06 PM rene_dev_: you cant really use opengl without glm :D
06:09 PM cerna: There is https://github.com/spotify/dh-virtualenv - not sure how well it works.
06:15 PM jepler: A sane way of doing 3d graphics in Python isn't in Debian? Sigh
06:16 PM rene_dev_: the lib is, the pyhton bindings are not
06:16 PM rene_dev_: its the lib you want for modern opengl
06:18 PM jepler: https://packages.debian.org/source/buster/pyopengl is no good?
06:19 PM rene_dev_: thats something completley different
06:19 PM rene_dev_: https://glm.g-truc.net/0.9.9/index.html
06:19 PM seb_kuzminsky: andypugh: keep in mind the etc/default/grub needs update-grub to "take effect", and if you boot to a different OS on a different partition you'll see a different /etc/default/grub
06:21 PM andypugh: I finally seem to have made it work by booting Mint, runnign update-grub from there, and choosing 2>5 to pick the 4th entry on the first menu.
06:21 PM andypugh: (it seems that some entries are not counted)
06:22 PM seb_kuzminsky: very odd, that has not been my experience
06:23 PM seb_kuzminsky: glad you got it working though
06:26 PM jepler: Can't numpy do all those things?
06:26 PM rene_dev_: yes, but its slow, and not designed to do this, and doesnt have the same nice helper functions
06:29 PM rene_dev_: I would like it if the gremlin would be fast, and could load megabytes of gcode without problems
06:29 PM andypugh: New minimal RTAI-crasher: https://pastebin.ubuntu.com/p/TZFZMD2yrf/
06:31 PM andypugh: (If I skip “loadrt threads” and put back “loadrt abs” then it’s all OK, or can do at least 3000 cycles anyway)
07:07 PM seb_kuzminsky: that's great andypugh
07:07 PM seb_kuzminsky: getting close to something that the rtai people can debug
07:37 PM andypugh: I have found something a bit odd in hal_lib
07:40 PM andypugh: https://github.com/LinuxCNC/linuxcnc/blob/master/src/hal/hal_lib.c#L3137
07:40 PM andypugh: sizeof(hal_thread_t) is 120 bytes
07:42 PM andypugh: which makes this code look a bit strange: https://github.com/LinuxCNC/linuxcnc/blob/master/src/hal/hal_lib.c#L2903
09:01 PM jepler: andypugh can you be more specific about what you're theorizing?
09:01 PM andypugh: I am not really theorizing. I am just not understanding why that alignment happens.
09:02 PM andypugh: I don’t think it’s the problem. That has to be somewhere in RTAI as Uspace works.
09:02 PM jepler: I guess we know some things need to be 16-aligned actually
09:02 PM jepler: The whole stack alignment thing
09:03 PM andypugh: Though I think this is different.
09:04 PM andypugh: Unfortunately the test routine that was crashing for me has stopped crashing…
09:05 PM jepler: Yes the consequence was visible immediately, not at shut down time, if I understand the problem right
09:06 PM andypugh: And that code is just aligning things inside the HAL shared memory.
09:09 PM andypugh: (I don’t think that there is any likelyhood that kernel pointers and userspace pointers to the same shared memory might be aligned differently, I think that would stop all HAL pins working properly)
09:10 PM andypugh: Currenlty I am trying to follow execution through RTAPI to RTAI. But it doesnt help that my test code has started to not-crash
09:10 PM andypugh: But, it’s 3am and I have a meeting at 0830…
09:11 PM jepler: 'night
09:11 PM jepler: you're young, you'll be fine on a few hours sleep :-P
09:35 PM -!- #linuxcnc-devel mode set to +v by ChanServ
09:45 PM seb_kuzminsky: sorry dgarr i broke the buildbot, your build is going to fail
09:46 PM skunkworks: nice
09:46 PM skunkworks: You have 1 job!
09:47 PM seb_kuzminsky: i forgot what directory the build step was running in
09:47 PM * seb_kuzminsky shrugs
09:47 PM seb_kuzminsky: you want 5 nines you gotta pay for food & healthcare
09:48 PM seb_kuzminsky: oh, i guess it's going to build after all
09:59 PM jepler: you want any nines at all ...
10:00 PM jepler: someone here makes a joke about offering "9 5s" of reliability but it's probably not original
10:00 PM jepler: 9.9999% is also not very good
10:05 PM skunkworks: wait - you do more than buildbot administation? Crazy ;)
10:07 PM seb_kuzminsky: lol, these days i barely do that
10:07 PM skunkworks: eh - we takes what we can gets..
10:08 PM seb_kuzminsky: those moodily lit slow-mo videos of you screwing and unscrewing your hex bore makes it all worth it :-)
10:08 PM seb_kuzminsky: jepler: i'm going to start using 9 fives
10:12 PM skunkworks: seb_kuzminsky: I was thinking of making a video of exactly that.. Mood lighting and everything...
10:12 PM skunkworks: Guess what I am doing right now?
10:12 PM skunkworks: ;)
10:14 PM skunkworks: I showed it to a guy at work and he didn't want to give it back.. I have to make him one now..
10:18 PM seb_kuzminsky: can you engrave "LinuxCNC.org" along one of the umm, curved flats?
10:21 PM skunkworks: you are going to make me buy a 4th/5th axis - arent you...
10:21 PM skunkworks: eh - maybe I could do it with just a 4th axis..
10:22 PM seb_kuzminsky: aww, i was hoping for just 3 and the spindle
10:25 PM skunkworks: Hmm.. Maybe I could stipple it as it is spinning...
10:25 PM skunkworks: jepler: !
10:27 PM skunkworks: It is funny how idea flow freely when not working 100% it is nice.
10:28 PM seb_kuzminsky: i agree
10:32 PM Tom_L: it's always good when this channel is busier than the main one :)
10:35 PM -!- #linuxcnc-devel mode set to +v by ChanServ
10:36 PM seb_kuzminsky: linuxcnc-build: force build --branch=master 0000.checkin
10:36 PM linuxcnc-build: build #6798 forced
10:36 PM linuxcnc-build: I'll give a shout when the build finishes
10:40 PM linuxcnc-build: build #1 of 1660.rip-buster-python3 is complete: Failure [4failed configuring] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1660.rip-buster-python3/builds/1
10:43 PM seb_kuzminsky: rene_dev_ jepler that's the new experimental python3 buildslave croaking because it can't find the packages it needs, because debian/configure didn't know it needed to install python3 stuff
10:43 PM seb_kuzminsky: i guess i'll install them by hand for now, and we can solve the debian part when we get to packaging later
10:57 PM skunkworks: Tom_L: here hear!
11:28 PM linuxcnc-build: build #6798 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/6798
11:29 PM seb_kuzminsky: linuxcnc-build: force build --branch=master 0000.checkin
11:29 PM linuxcnc-build: build #6799 forced
11:29 PM linuxcnc-build: I'll give a shout when the build finishes
11:47 PM linuxcnc-build: build #2 of 1660.rip-buster-python3 is complete: Failure [4failed compile] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1660.rip-buster-python3/builds/2