#linuxcnc-devel Logs

Oct 05 2023

12:12 AM pere: Is Mark van Doesburg still around in the project?
01:11 AM linuxcnc-build2: Worker `checkin` is missing. It was seen last on Wed Oct 4 23:08:50 2023.
04:54 AM andypugh: seb_kuzminsky: Looking at the linuxcnc-gbp recipe, I don’t understand how the .deb get to debian.
05:31 AM andypugh: Who signed the current package?
06:44 AM pere: andypugh: normally I sign the new upload.
06:45 AM andypugh: Hi. I have just realised that I was up till 0430 last night trying to get a clean biuild on the buildbots for no reason. The debian point release deadline was _last_ weekend
06:46 AM pere: andypugh: more point releases will follow.
06:46 AM andypugh: Yeah, it’s not all bad.
06:46 AM pere: sad to hear. do not pay attention to those deadlines myself, so I did not notice when you said so.
06:47 AM andypugh: The email said “7th Oct” which was what stuck in my mind, and not the “uplads will close the weekend before”
06:48 AM pere: anyway, <URL: http://www.hungry.com/~pere/debian-sponsoring.html > are my preferences. sponsoring is the process when someone with upload rights sign and upload packages created by someone without upload rights.
06:48 AM andypugh: Though, because it was 0430 and i was 3/4 asleep I messed up the release commit to linuxcnc and didn’t commit the tag properly.
06:49 AM andypugh: So I am not sure if the buildbot will do its thing.
06:51 AM andypugh: It probably wasn’t going to happen anyway, I need to leave to fly to Stuttgart in a couple of hours. (For the LinuxCNC fest)
06:52 AM Tom_L: have a safe trip and thanks for your efforts :)
06:55 AM andypugh: It’s possible that I might go straight to a 2.9.1 release on Sunday instead now. It would be nice to squash the G72 bugs
06:56 AM pere: andypugh: I emailed Mark van Doesburg, the original author of the G7X code, to see if he had any ideas. No idea if he got the email or is still interested, but thought it best to give it a go.
06:57 AM andypugh: I wish I had kept my version instead. But he seemed keen, and mine wasn’t finished. Mine had much better docs, though.
06:59 AM rmu: it should be possible to fix it if this sign-flipping is the only problem
06:59 AM pere: wish we had a test set for it, to verify that it actually work, or to demonstrate its weaknesses, to ensure it improve over time. :)
07:00 AM andypugh: It’s not trivial to test without some tricky machine-path parsing
07:01 AM rmu: doing some checks on rs274 output like initial/end position and number of "dives" would be a start
07:06 AM andypugh: For a specific input profile, yes. Though pere has pretty much done exactly that
07:07 AM pere: not really. I have just parroted code in bug reports. I have no idea what the behaviour should be, only what it should not be. :)
07:08 AM pere: andypugh: why is tricky machine-path parsing needed?
07:12 AM andypugh: To be sre that the path output by the command matches what is required
07:12 AM pere: would not controlling the input be enough?
07:13 AM andypugh: The problem is that some inputs give the wrong output. How do we check for the right output?
07:13 AM andypugh: It’s a very complicated canned cycle
07:18 AM pere: My approach has been to get it right, run it once, move results to expected. This way we at least discover when the result change, even if it was originally incorrect.
07:20 AM rmu: i think what pere says, basically tests that will only catch regressions / newly introduced bugs
07:22 AM rmu: I wonder if these canned cycles are really that useful. 50ies CAM integrated into G-Code.
07:28 AM -!- #linuxcnc-devel mode set to +v by ChanServ
08:03 AM pere: Mark van Doesburg responded, was not aware of the issues and will have a look tonight.
08:29 AM rene-dev5: Rmu will you be in Stuttgart?
08:39 AM rmu: yes
08:40 AM rmu: i will bring two half-finished stmbl clones with ST igbt module, some other stuff, and try to make progress on the bezier-acceleration-stuff
08:46 AM andypugh: pere: Can you look at https://github.com/LinuxCNC/linuxcnc/tree/master/debian and possibly work out why those instructions don’t work? Lots of “gbp:error: You are not on branch 'master' but on 'debian/unstable’ and “Non-native package 'linuxcnc' has invalid version '2.9.0+git20231005.844045edfb’” and “upstream/2.9.0 is not a valid treeish”
08:48 AM andypugh: That said, I think that I managed to create a .dsc but can’t sign it. (Well, I _could_ sign it, but not with the right key, I expect)
08:53 AM pere: you are building using gbp-buildpackage?
08:59 AM andypugh: That stage of the recipe is a test build in a docker container with gbp
09:01 AM andypugh: But I got the same messages with dpkg-buildpackage. Though by that point I had suffixed the version string with a -1 to make it look “non-native” and learned to use the —debian-branch=debain/unstable but the latter seems wrong.
09:02 AM pere: I suspect debian/gbp.conf might ned to be adjusted.
09:02 AM andypugh: Might need to be invented :-)
09:03 AM andypugh: The file is absent https://github.com/LinuxCNC/linuxcnc/tree/master/debian
09:03 AM pere: ah.
09:03 AM pere: it can contain the branch to build from.
09:04 AM pere: andypugh: I believe seb_kuzminsky built from a linuxcnc-gbp repo with debian specific adjustments, if I remember correctly.
09:05 AM andypugh: Bear with me. I was in the wrong repo
09:05 AM andypugh: It’s also missing from there: https://github.com/LinuxCNC/linuxcnc-gbp/tree/debian/unstable/debian
09:06 AM pere: My dirty git repo seem to build with gbp buildpackage here. not done yet.
09:07 AM andypugh: What _should_ the version string in the changelog be? 2.9.0 ? 1:2.9.0 ? 1:2.9.0-1 ? 2.9.0+git20231005.844045edfb-1 ?
09:23 AM pere: ah, that mess. andypugh, are you uploading to debian? <URL: https://tracker.debian.org/linuxcnc/ > give the authorative number.
09:24 AM pere: 2.9.0~pre1+git20230208.f1270d6ed7-1 was the last upload. If you are uploading 2.9, I guess 2.9.0-1 would be a good next number.
09:25 AM pere: the epoc part (1:) is only in the non-Debian repo packages, and introducing such epoc need to be coordinated within Debian, so should not be done now.
09:26 AM pere: ~pre1... is just to indicate that this is before 2.9.0. ~ is less than any other character in the version string.
09:26 AM pere: quite a lot of lintian warnings about the package. my gbp buildpackage succeeded a minute ago.
09:54 AM andypugh: I guess the git hash is just so that there is a way to identify the exact place in time
09:55 AM andypugh: seb_kuzminsky: Is it possible to force the buiuldbots to build release debs? It was way to late at night, and, though I added the tag, I failed to push it with the last commit.
10:41 AM pere: andypugh: yes, hash is useful when there is no tag.
10:43 AM pere: andypugh: if you add yourself as uploader in debian/control, you get your own page like <URL: https://qa.debian.org/developer.php?email=jepler%40gmail.com > with lots of useful links.
11:47 AM seb_kuzminsky: andypugh: i think it should be enough to trigger a rebuild of the release commit, now that the release tag is in the repo
11:48 AM seb_kuzminsky: i triggered that build a little while ago and it's nearly at the "build release debs" part so we'll soon know if it's gonna work the easy way or if i'm going to have to debug it
12:08 PM seb_kuzminsky: nope, it noticed that the 2.9 commit was already in its local repo so it didn't re-fetch, so it still doesn't have the v2.9.0 tag
12:08 PM seb_kuzminsky: err....
12:10 PM seb_kuzminsky: linuxcnc-build: force build --revision=v2.9.0 0000.checkin
12:10 PM seb_kuzminsky: maybe?
12:10 PM linuxcnc-build: The build has been queued, I'll give a shout when it starts
12:21 PM seb_kuzminsky: linuxcnc-build2: force build --revision=v2.9.0 00-checkin
12:21 PM linuxcnc-build2: Force build successfully requested.
12:31 PM seb_kuzminsky: ugh, one of the servers that's running the buildbot VMs is failing
12:34 PM -!- #linuxcnc-devel mode set to +v by ChanServ
01:06 PM seb_kuzminsky: linuxcnc-build2: force build --revision=v2.9.0 00-checkin
01:06 PM linuxcnc-build2: Force build successfully requested.
01:25 PM seb_kuzminsky: linuxcnc-build2: force build --revision=v2.9.0 00-checkin
01:25 PM linuxcnc-build2: Force build successfully requested.
01:27 PM linuxcnc-build2: Build [#1614](http://buildbot2.highlab.com/buildbot/#builders/19/builds/1614) of `10-rip.debian-08-jessie-i386` 3completed successfully.
01:27 PM linuxcnc-build2: Build [#1613](http://buildbot2.highlab.com/buildbot/#builders/10/builds/1613) of `10-rip.ubuntu-12-04-precise-rtai-i386` 3completed successfully.
01:27 PM linuxcnc-build2: Build [#1612](http://buildbot2.highlab.com/buildbot/#builders/22/builds/1612) of `10-rip.debian-09-stretch-rtpreempt-i386` 3completed successfully.
01:27 PM linuxcnc-build2: Build [#1612](http://buildbot2.highlab.com/buildbot/#builders/12/builds/1612) of `10-rip.debian-07-wheezy-amd64` 3completed successfully.
01:29 PM linuxcnc-build: build forced [ETA 3h11m13s]
01:29 PM linuxcnc-build: I'll give a shout when the build finishes
01:51 PM linuxcnc-build2: Worker `checkin` is missing. It was seen last on Thu Oct 5 11:48:26 2023.
01:53 PM linuxcnc-build2: Build [#1607](http://buildbot2.highlab.com/buildbot/#builders/14/builds/1607) of `10-rip.debian-10-buster-rtpreempt-amd64` 8completed with warnings.
01:55 PM linuxcnc-build2: Build [#1608](http://buildbot2.highlab.com/buildbot/#builders/7/builds/1608) of `10-rip.debian-11-bullseye-rtpreempt-amd64` 8completed with warnings.
02:21 PM linuxcnc-build2: Build [#1612](http://buildbot2.highlab.com/buildbot/#builders/11/builds/1612) of `10-rip.debian-12-bookworm-rtpreempt-amd64` 8completed with warnings.
02:37 PM rigid: rene-dev5: i think you broke compile with --enable-toolnml
02:38 PM rigid: i get "/home/pi/linuxcnc-dev/src/emc/task/taskclass.cc:147: undefined reference to `tool_mmap_creator(EMC_TOOL_STAT const*, int)' collect2: error: ld returned 1 exit status" with recent master
02:40 PM rigid: it's only included with -DTOOL_NML (TOOL_NML activated by ./configure --enable-toolnml)
02:42 PM rigid: wrapping the relevant code with "#ifndef TOOL_NML ..." should fix it
02:48 PM rigid: andypugh: quick update regarding linuxcncrsh: i found a lot more bugs but I'm still struggling to even get it to work. the only command I tried that actually works is "get time". I'm currently trying to get a fresh compile and running the tests.
03:43 PM linuxcnc-build: build #3165 of 3304.dsc-buster-rtpreempt is complete: Failure [4failed update changelog] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/3304.dsc-buster-rtpreempt/builds/3165
04:38 PM Unterhaus_ is now known as Unterhausen
04:45 PM linuxcnc-build: build #10315 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/10315
04:49 PM pere: andypugh rmu: Proposed fix for the g72 issue, see <URL: https://github.com/LinuxCNC/linuxcnc/pull/2677 >. Got the patch from the original author, who said he could no longer push to github.
04:59 PM linuxcnc-build2: Build [#1401](http://buildbot2.highlab.com/buildbot/#builders/21/builds/1401) of `20-docs` 6stopped with exception.
05:46 PM seb_kuzminsky: linuxcnc-build: force build --branch=2.9 --revision=v2.9.0 0000.checkin
05:46 PM linuxcnc-build: build forced [ETA 3h11m13s]
05:46 PM linuxcnc-build: I'll give a shout when the build finishes
05:54 PM rene-dev5: Rigid what’s the use case for that?
05:56 PM linuxcnc-build2: Worker `checkin` is missing. It was seen last on Thu Oct 5 15:53:48 2023.
05:58 PM linuxcnc-build2: Build [#1402](http://buildbot2.highlab.com/buildbot/#builders/21/builds/1402) of `20-docs` 6stopped with exception.
05:58 PM linuxcnc-build2: Build [#1686](http://buildbot2.highlab.com/buildbot/#builders/13/builds/1686) of `00-checkin` 6stopped with exception.
06:27 PM rigid: rene-dev5: using nml for tooldata instead of mmap
06:29 PM rigid: i suppose the default mmap mode fails if tool channels are connecting different hosts.
07:55 PM linuxcnc-build: Hey! build 0000.checkin #10316 is complete: Success [3build successful]
07:55 PM linuxcnc-build: Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/10316
08:02 PM rene-dev5: Did you ever get that working
08:02 PM rene-dev5: different hosts
08:03 PM rene-dev5: The hal pins of the UI don’t work
08:52 PM linuxcnc-build: build #3018 of 4041.deb-buster-rtpreempt-amd64 is complete: Failure [4failed shell_3 shell_4] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4041.deb-buster-rtpreempt-amd64/builds/3018
09:29 PM linuxcnc-build: build #2909 of 4042.deb-buster-rtpreempt-rpi4 is complete: Failure [4failed shell_3 shell_4] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4042.deb-buster-rtpreempt-rpi4/builds/2909