#linuxcnc-devel Logs

Sep 20 2019

08:41 AM JT-Shop: because there is no axis.0...
08:45 AM JT-Shop: 2.8 and 2.9 have not eliminated the axes...
09:19 AM Tom_L: but it shows up in the list
09:19 AM JT-Shop: no it does not... read the pin name
09:20 AM JT-Shop: that's why you should not guess at pin names and print out a list of pins for that configuration
09:20 AM JT-Shop: axes can be only abcuvwxyz not 0
07:53 PM andypugh: Does anyone mind if I try something to see if it breaks the buildbot?
07:55 PM andypugh: I want to make a change to the rtai build, and the buildboit seems like an easier way to check what it breaks than setting up a whole load of machines myself.
08:05 PM seb_kuzminsky: andypugh: that's a totally valid way to use the buildbot
08:05 PM seb_kuzminsky: just do your experiments on a personal test-branch rather than one of the long-lived branches, of course
08:07 PM andypugh: Ah, yes, that would have been the way to do it.
08:07 PM andypugh: (10 seconds too late)
08:08 PM andypugh: I think it might well be OK, it’s changing an include from /asm/ to /linux/
08:12 PM andypugh: Currently RTAPI does not compile on 4.14 so the test is really whether it is backwartds compatible. And if it isn’t I have no real idea how to handle that.
08:12 PM andypugh: https://github.com/LinuxCNC/linuxcnc/commit/0dbbfb3ae106feea27dab4753ae40c21b6ea06d3
08:12 PM seb_kuzminsky: you could use the kernel version check macros to include different files on different kernel version
08:12 PM seb_kuzminsky: *versions
08:14 PM andypugh: I am sure _you_ could :-)
08:15 PM andypugh: Oh, and while you have popped your head out of your burrow, could you look at https://github.com/LinuxCNC/linuxcnc/issues/609 ?
08:16 PM andypugh: I can do the work, but I just want to be sure I am right..
08:18 PM seb_kuzminsky: https://github.com/LinuxCNC/linuxcnc/blob/master/src/rtapi/rtapi_proc.h#L77
08:19 PM andypugh: commas not dots? Now that’s an interesting “gotcha”
08:20 PM seb_kuzminsky: it's not a string, it's three integer arguments, separated by commas as usual
08:20 PM seb_kuzminsky: did i really break jogging, back in 2015?
08:20 PM andypugh: Well, there’s the thing. It seems unlikely that nobody would have noticed.
08:21 PM andypugh: Hence my query and not jumping in with a “fix”
08:21 PM seb_kuzminsky: foolish of me not to write a test...
08:28 PM seb_kuzminsky: i build 2.7, started sim axis, and poked at the halui pins manually, using halcmd
08:28 PM seb_kuzminsky: everything seems to work like it should
08:29 PM seb_kuzminsky: if i select joint.0, set halui.jog-speed to a non-zero value, and set halui.jog.selected.plus, then joint 0 (X) jogs in the positive direction
08:29 PM seb_kuzminsky: if i then select joint.1, X stops jogging and Y starts jogging
08:29 PM seb_kuzminsky: that's all as it should be
08:34 PM seb_kuzminsky: oh, you're on 2.8, post JA
08:34 PM seb_kuzminsky: i tested the wrong thing
08:34 PM seb_kuzminsky: hold on
08:44 PM andypugh: I don’t actually expect a difference,
08:44 PM andypugh: I think it probably does work. But looking at the code it shouldn’t :-)
08:45 PM seb_kuzminsky: joint jogging via halui's selected-joint mechanism works in 2.8, just like it did in 2.7
08:47 PM andypugh: Aye. As I said, folk would have noticed.
08:47 PM seb_kuzminsky: axis jogging is slightly broken
08:47 PM andypugh: Does your test setup set num_axes < MAX_AXES?
08:48 PM seb_kuzminsky: i have [TRAJ]COORDINATES = X Y Z and [KINS]JOINTS = 3
08:49 PM seb_kuzminsky: but halui shows 9 axes (xyzabcuvw)
08:50 PM seb_kuzminsky: but only 3 joints (012)
08:50 PM seb_kuzminsky: so maybe halui always has all 9 axes, no matter what the ini says, and that's why it works?
08:50 PM andypugh: That would explain it.
08:51 PM seb_kuzminsky: ¯\_(ツ)_/¯
08:51 PM andypugh: If you are confident it works, let’s close it.
08:51 PM seb_kuzminsky: 🤷
08:51 PM seb_kuzminsky: fine by me!
08:51 PM seb_kuzminsky: oh, there is a small bug with axis jogging (but not joint jogging)
08:52 PM seb_kuzminsky: if you have one axis selected and hold down halui.axis.selected.plus, then that axis jogs like it should
08:52 PM seb_kuzminsky: but if you then select a different axis (while holding .selected.plus true) then the first axis stops and the second axis doesn't start
08:52 PM seb_kuzminsky: the second axis should start
08:53 PM andypugh: At least it’s a safe bug
08:53 PM seb_kuzminsky: (at least, that's what the joint jogging does in 2.7 and 2.8)
08:53 PM andypugh: I can even persuade myself that it is a desirable behaviour
08:54 PM andypugh: Did you see the purple column in the waterfall? Seems to have halted everything.
08:58 PM seb_kuzminsky: i'll go kick it again
08:59 PM seb_kuzminsky: it's back