#linuxcnc-devel Logs

Jul 19 2024

10:21 AM pere: wow, <URL: https://news.slashdot.org/story/24/07/18/2050235/nvidia-will-fully-transition-to-open-source-gpu-kernel-modules-with-r560-drivers >.
10:37 AM -!- #linuxcnc-devel mode set to +v by ChanServ
11:59 AM rmu: pere: you mean the rt-preempt patches?
12:03 PM rmu: from last year https://lwn.net/Articles/938236/
12:08 PM mrec: opensource kernel modules because they can put their interesting part into closed source userspace parts
12:09 PM mrec: so not a real "wow"
12:09 PM mrec: having so many kernel drivers is absolute nonsense anyway
12:09 PM mrec: even though linux is monolithic they have their hooks for userspace drivers and they could have specified much earlier too
12:10 PM Unterhaus_ is now known as Unterhausen
12:47 PM andypugh_: pere: Possibly, though I am not sure how to use it.
01:20 PM Unterhaus_ is now known as Unterhausen
02:03 PM pere: andypugh_: it can be compared to your result, to see if we end up with the same result?
03:06 PM andypugh_: I think I have done the merge now. Not sure how to compare branches?
03:07 PM andypugh_: git checkout —ours was the right choice in many cases.
03:10 PM pere: andypugh_: this will fetch and compare my branch to your branch: git remote add petterreinholdtsen htts://github.com/petterreinholdtsen/linuxcnc.git; git fetch petterreinholdtsen; git diff andybranch petterreinholdtsen/master-2.9-merge-test-2024-07-18
03:11 PM pere: s/htts/https/
03:21 PM andypugh_: Hmmm. Currently failing to compile.
03:21 PM andypugh_: https://gist.github.com/andypugh/b9cbded9fc7687b3f350c57f0a29bad9
03:22 PM pere: andypugh_: did you notice my failing build due to a test change?
03:23 PM andypugh_: Not yet. All builds have been failing for some time.
03:23 PM pere: un unstable, yes
03:23 PM andypugh_: That’s on the list
03:24 PM pere: andypugh_: see for example <URL: https://github.com/petterreinholdtsen/linuxcnc/actions/runs/10006040847/job/27657930764 > for the test failure.
03:25 PM andypugh_: Did you look at the diff?
03:25 PM andypugh_: I can only think about one thing at a time.
03:25 PM pere: yes, but not sure what I was to use it for.
03:25 PM andypugh_: Just pointing out that we got different results. Which seems a bit unexpected.
03:26 PM pere: ah, so this is the diff between your 2.9->master merge and my master-2.9-merge-test-2024-07-18?
03:27 PM pere: Most of it seem to be simply missing my fixes to the 2.9 patches to adapt them to master.
03:29 PM andypugh_: Yes, I only tried to resolve merge conflicts.
03:29 PM pere: andypugh_: the debian/linuxcnc.install.in compencate for the 2.9 branch failing to include the new manual pages in the debian package. this is a fatal error when building with debhelper 13, so had to be fixed.
03:29 PM pere: the debian/changelog change seem to be someone replacing UNRELEASED at the appropriate place. this was not in 2.9 when I merged, which surprised me but I ignored.
03:30 PM pere: The lib/python/qtvcp/widgets/versa_probe.py change seem to be removal of trailing space, no recollection about doing anything there. did you loose a white space change?
03:31 PM pere: The src/emc/rs274ngc/interp_g7x.cc compencate for the 2.9 change being correct in 2.9, but incorrect in master, as 2.9 uses int for cutter_comp_side, while master uses a C++ class.
03:31 PM andypugh_: Possibly. Though the only whitespace thing that I did consciously was checkut —ours of emcrsh.cc then manually take out the trailing spaces that were causing the merge conflict, having checked that there were no functional changes.
03:32 PM pere: In the same file, I suspect the path.front is wrong and should be paths.front in master.
03:32 PM pere: what does -ours do here?
03:32 PM andypugh_: interp_g7x is the headache. I might have to check it out from 2.9 then cherry-pick the master chanhes.
03:33 PM andypugh_: —ours / --theirs is a lazy way to settle a merge conflict by settling for one or the other unchanged files.
03:33 PM pere: The src/emc/task/emctask.cc change is simply me applying the strncpy->rtapi_strlcpy migration on new code in master, to ensure the change from 2.9 is properly applied in master, also with new code.
03:34 PM pere: is -ours = master and -theirs = 2.9?
03:34 PM andypugh_: depends on what is being merged. If you are in branch A and merge B then A is ours and B is theirs
03:35 PM pere: The src/emc/usr_intf/emcrsh.cc changes are me ensuring all changes done in 2.9 were also applied in master, ie adding the commend and removing the same trailing space and blank lines as in 2.9.
03:35 PM pere: That conclude my summary of differences between your merge and mine. any questions?
03:36 PM pere: I guess the install.in change really should go into 2.9 too, and be merged forward, instead of the way I did it. the rest I suspect should be in the merge commit.
03:37 PM pere: Is your merge edition of src/emc/rs274ngc/interp_g7x.cc building? I doubt it.
03:37 PM andypugh_: No, it’s not. I might just grab your versio if your does.
03:38 PM andypugh_: I guess 2.9.3 in the changelog should actually be “unstable” rather than UNRELEASED, but 2.9.2 should be UNRELEASED?
03:38 PM pere: 0b80d453da34490e8a244b8d5054015b53011457 show the change I applied to master. no changes to the lines above and below.
03:40 PM pere: andypugh_: the UNRELEASED/unstable/stable/etc markers are used in debian for the archive handling system to know what to do with a uploaded package. for linuxcnc, I guess it could be most sensibly used to indicate that the package is stamped and released, so no changes should be done to the changelog block.
03:43 PM pere: but the 2.9 merge changed how G71 behave, and I have no idea if this is an intended change or a bug.
03:43 PM andypugh_: I would need to ask the author.
03:45 PM andypugh_: I am unclear on whether the new man1 pages should be in master now or not.
03:46 PM pere: andypugh_: as soon as the groff files show up in master, I will convert them to adoc in master.
04:18 PM linuxcnc-build2: Worker `checkin` is missing. It was seen last on Fri Jul 19 14:12:34 2024.
04:18 PM linuxcnc-build2: Worker `jessie-i386` is missing. It was seen last on Fri Jul 19 14:12:35 2024.
04:22 PM andypugh_: Not sure we care about Jessie-i386, as cruel as that might seem.
04:26 PM linuxcnc-build2: Worker `bookworm-rtpreempt-amd64` is missing. It was seen last on Fri Jul 19 14:20:10 2024.
04:33 PM andypugh_: pere: I took your interp_g7x.cc and it compiles. But the g71-endless-loop2 test fails.
04:33 PM andypugh_: That same test passes in master _and_ in 2.9.
04:46 PM andypugh_: Actually, endless_loop2 isn’t in 2.9
05:27 PM lcnc-relay: <captainhindsight_.#0> andypugh_ : are we ready for RTAI against your 2.9.3 release?
05:28 PM lcnc-relay: <andypugh#0> I am niot sure that I 100% understand the question. I already built and published rtai debs for 2.9.3
05:28 PM lcnc-relay: <andypugh#0> * not
05:28 PM lcnc-relay: <captainhindsight_.#0> oh you built your own against it
05:29 PM lcnc-relay: <captainhindsight_.#0> I didn't see
05:29 PM lcnc-relay: <andypugh#0> But you need to use a sub-branch andypugh/2.9-rtai as the main debian config folder doesn't work for RTAI.
05:30 PM lcnc-relay: <andypugh#0> I build those debs on the day of release, because my test machines all have rtai.
05:30 PM lcnc-relay: <captainhindsight_.#0> I see the RTAI 5.4.258
05:32 PM lcnc-relay: <captainhindsight_.#0> memleak had updated to https://onelaboratories.us/abc/ 5.4.279
08:19 PM mrec: when stopping axis it seems to unset the key repeat rate? Is anything known in that area
08:19 PM mrec: (latest git linuxcnc)
08:20 PM mrec: 2.10pre
08:20 PM mrec: two ways to set it back... switch to a local tty terminal and back to xorg or set it with xset
09:26 PM linuxcnc-build: build #3396 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/3396 blamelist: fsabbatini89 <fsabbatini@gmail.com>, Petter Reinholdtsen <pere@hungry.com>, andypugh <andy@bodgesoc.org>, Mark <mark.vandoesburg@hetnet.nl>
09:28 PM linuxcnc-build: build #10757 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/10757 blamelist: fsabbatini89 <fsabbatini@gmail.com>, Petter Reinholdtsen <pere@hungry.com>, andypugh <andy@bodgesoc.org>, Mark <mark.vandoesburg@hetnet.nl>
11:20 PM pere: andypugh_: so should tests/interp/g71-endless-loop2 be added to 2.9?
11:21 PM pere: andypugh_: tests/interp/g71-with-g70 is also missing in 2.9.