#linuxcnc-devel Logs

Oct 19 2018

#linuxcnc-devel Calendar

05:08 AM CMorley: andypugh:https://forum.linuxcnc.org/24-hal-components/35391-program-stops-executing-after-spindle-at-speed-reached#119067
06:03 AM jthornton: is that the same problem Loetmichal had yesterday on #linuxcnc?
03:51 PM jepler: forum update, yay/ugh
03:52 PM jepler: and we should be back up after that brief interruption
03:53 PM dgarr: linuxcnc-build: force build --branch=dgarr/external_offsets 0000.checkin
05:11 PM andypugh: mozmck: What do you think? Merge external_offsets (well tested) and reverse-run (less well tested but harder to activate by accident) and aim to release by new-yea?
05:12 PM jthornton: yippie!
05:12 PM andypugh: If I knew what the release manager did I would offer to take over :-)
05:19 PM dgarr: linuxcnc-build: force build --branch=master 0000.checkin
05:21 PM andypugh: Hi dgarr
05:22 PM andypugh: I tried cherry picking that single commit last night, on top of master. Built cleanly and passed all tests.
05:22 PM andypugh: I am not sure the actual commit needs much more work, unless you have tweaked it.
05:24 PM andypugh: What I haven’t quit managed to do yet is put https://github.com/LinuxCNC/linuxcnc/pull/512 on top of master + ext_off (which a guy on the forum seems to need(
05:26 PM dgarr: i don't understand the runtests failure as i cannot replicate, suspect problem using linuxcncrsh on the single buildbot machine: rip-jessie-rtprempt-amd64
05:27 PM dgarr: ref: http://buildbot.linuxcnc.org/buildbot/grid
05:27 PM dgarr: so testing latest master build to see if runtests fail repeats on rip-jessie-rtpreempt-amd64
05:28 PM andypugh: runtests defintely passes on my local stretch-uspace
05:28 PM andypugh: Having read what you said more carefully, I now understand what you are trying to do
05:29 PM dgarr: the commit b546cfe01135 passed on all buildbot machines, the commit 784db03355d5 only changed the commit message yet failed on one machine
05:31 PM andypugh: Is there a way to see what happened? Locally there is a .failed file (I think) but I don’t know if we can get there on the buldbot
05:32 PM dgarr: you can find the log using the buildbot grid: http://buildbot.linuxcnc.org/buildbot/builders/1530.rip-jessie-rtpreempt-amd64/builds/2265/steps/runtests/logs/stdio
05:33 PM dgarr: the first failure (the log references are now gone) had one odd/unexpected failure, the log above (on the repeat build) has multiple failures with many of them showing timeout using linuxcncrsh -- my guess is there is some "bad" system state due to original linuxcncrsh timeout or related problem
05:34 PM dgarr: if failures repeat on the current master branch build, i would suggest rebooting therip-jessie-rtpreempt-amd64 machine
05:36 PM dgarr: expected completion (buildbot waterfall) shows 2hrs21minutes or 18:55 mdt
06:10 PM dgarr: found log page for the failing machine: http://buildbot.linuxcnc.org/buildbot/builders/1530.rip-jessie-rtpreempt-amd64 It shows rip-jessie-rtpreempt-amd64 runtests fails on last three builds including latest rerun of master at commit 0cab36566e8d -- so i think the problem is with the buildbot machine
06:11 PM rene_dev_: andypugh can you confirm that PR 512 works?
06:17 PM dgarr: The first failing build (build #2264) for rip-jessie-rtpreempt-amd64 occurred after running 169 tests (out of 224) with messages "USRMOT: ERROR : command timeout" ref: http://buildbot.linuxcnc.org/buildbot/builders/1530.rip-jessie-rtpreempt-amd64/builds/2264/steps/runtests/logs/stdio
06:27 PM andypugh: rene_dev_: I can’t. I am trying to find a way to create a patch for it on top of master/external_offsets for a guy on the forum running that software with a machine that shows the problem but who is unfamilar with git (but is runing a RIP that Dewey helped him make)
06:27 PM rene_dev_: just push everything into master :D
06:28 PM rene_dev_: but I guess cherrypicking is what you are looking for
06:28 PM andypugh: I got as far as https://patch-diff.githubusercontent.com/raw/LinuxCNC/linuxcnc/pull/512.diff by manually editing URLs
06:29 PM andypugh: But that patch doesn’t take on master (got won’t even try, vanilla patch has a minor issue that I need to margetool)
06:29 PM andypugh: (git won’t evem try)
07:08 PM hazzy-m: I think something must be wrong with the bot that builds the docs, the python interface docs have not been updated since 2018-10-18 22:49:08 MDT ...
07:55 PM andypugh: dgarr: Moses said “if there are no stong objections” and it’s been _hours_ witout any :-)
07:57 PM andypugh: The way I see it is that external-offsets has had _no_ error reports for 18 months with at least 6 users that I know of (and the guy complaining about a TP bug was a new one who mafe it seven)
08:01 PM andypugh: So given that Multispindle is already merged and raised 3 bugs in a week, I think that running EO in parallel with waiting for MS so settle down makes sense. This is mainly because my incompetence seems to have introduced a “wait and see” period that other changes might as well join in on.
08:22 PM rene_dev_: andypugh yeah, why not merge it. I agree
08:24 PM hazzy-m: it's one of the best tested feature branches. Go for it!
08:27 PM hazzy-m: I think maybe it would have been best to have tagged/released 2.8 when it was extremely stable a few weeks ago, but now might as well take advantage to merge what we can before the release
08:50 PM dgarr: seb_kuzminsky: when you have a chance, could you look at the buildbot slave: rip-jessie-preemptrt-amd64 -- the runtest failures seem to indicate a problem with it alone -- maybe it needs a reboot . more info in channel logs^^
10:20 PM mozmck: The only thing that bothers me a little reading the docs for external offsets is that there is no deceleration if external offsets are enabled and you hit a soft limit either while jogging or during a coordinated move.
10:23 PM mozmck: But I guess that is probably not much of an issue as long as you know about it and are careful.
10:23 PM andypugh: It might not be ideal, but if EO is not enabled (default) then that won’t be seen, and no coordinated move should hit a soft limit anyway?
10:23 PM mozmck: true.
10:23 PM andypugh: Anyway, 0422 here,,