#linuxcnc-devel Logs

Jun 12 2019

#linuxcnc-devel Calendar

08:36 AM jthornton: hmmm same error as before
08:39 AM jthornton: linuxcnc-build: force build --branch=2.8 0000.checkin
08:39 AM linuxcnc-build: build forced [ETA 1h18m05s]
08:39 AM linuxcnc-build: I'll give a shout when the build finishes
09:10 AM seb_kuzminsky: jthornton: i wrote up some debugging steps up above
10:39 AM linuxcnc-build: Hey! build 0000.checkin #5985 is complete: Success [3build successful]
10:39 AM linuxcnc-build: Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/5985
11:14 AM jthornton: seb_kuzminsky: thanks, I missed them in all the buildbot posts but Tom_L let me know
03:44 PM cradek: I think it would be a regression to make feed override affect jogs again
03:44 PM cradek: I see it like this: we have approximately 4 kinds of motion: feeds, rapids, jogs, homing
03:45 PM cradek: each of the first 3 has its own control in the gui, one control per type of motion, the 4th (homing) doesn't and that is good because homing speeds should not be overridden
03:45 PM cradek: having one control change two motion types, or worse, having two controls affect one type (you multiply them together in your head!?) violates the principle of least surprise
03:45 PM jthornton: I agree, if one wants feed and jog override to be connected they should do that in the GUI
03:46 PM cradek: I don't understand why someone would want both FO and jog speed to affect jogs (but I'd sure be willing to listen)
03:46 PM cradek: emc1 did that and we fixed it, and we might have gone back and forth a few times, I don't remember for sure
03:47 PM cradek: in emc1 FO affected everything - feeds, rapids, homing, jogs, and we separated them on purpose
03:47 PM Tom_L: as they should be
03:52 PM cradek: (I should've typed that into an email instead)
03:56 PM Tom_L: andypugh, nice video
03:56 PM Tom_L: wrong channel too :)
03:57 PM andypugh: cradek: He likes to hold down a keyboard button then adjust feed speed with 1, 2, 3 etc. when he can’t be bothered with G-code.
03:58 PM andypugh: I think that he might be mollified if shown how to do it in HAL.
03:59 PM cradek: I thought the jog speed adjusting buttons , and . worked that way too?
04:00 PM cradek: either way this is a gui issue, I agree keyboard jogging should be as good as possible
04:00 PM cradek: I can't run sim right now to try it, I'm in a restaurant on my phone.
04:01 PM andypugh: Ahm so you can’t check my current worry that halui jogging is broken in master?
04:01 PM cradek: nope I can't check anything, but I'll opine about anything you want
04:02 PM andypugh: Can you opine on code, or is that asking too much on a phone?
04:02 PM cradek: I can try
04:03 PM andypugh: halui,cc line 2001
04:04 PM andypugh: It seems to use num_axes where EMC_MAX_AXIS was intended.
04:04 PM andypugh: (that being the slot used for halui.jog-selected)
04:05 PM cradek: you mean axis_num?
04:07 PM andypugh: It’s axis_num in 2000 and num_axes in 2001
04:07 PM cradek: right they should all be axis_num (the loop counter) right?
04:08 PM andypugh: Which I think is deliberate but possibly wrong,
04:08 PM andypugh: I think 2001 is trying to look at the jog-selected pin
04:09 PM andypugh: retval = hal_pin_bit_newf(HAL_IN, &(halui_data->ajog_plus[EMCMOT_MAX_AXIS]), comp_id, "halui.axis.selected.plus");
04:11 PM cradek: the screwedup indentation in that blocks help make it clear as mud
04:12 PM cradek: I don't get what aselect_changed is supposed to do, but I think this is a simple loop over the axes with counter axis_num (should've used "i")
04:12 PM andypugh: I think that the “num_axes” ought to be “EMCMOT_NUM_AXES” as num_axes is nearly always < 9 (it’s based on [TRAJ]COORDINATES
04:13 PM cradek: oh and the axis_mask test will short-circuit any of the 9 that aren't defined on this machine?
04:13 PM cradek: you might be right (so that's two bugs)
04:13 PM andypugh: In any case, in trying to test whether it really is wrong I have managed to determine that the sim-axis-halui-pyvcp simulator can jog in workd mode, but not after homing.
04:14 PM cradek: hmm that smells backwards
04:15 PM andypugh: it’s driving the axis.L.jog pins properly, but nothing happens in 2.8 or master (but does in 2.7)
04:15 PM cradek: 2.7 is pre-ja isn't it?
04:15 PM andypugh: Yes
04:15 PM andypugh: So I might be missing an _axis_ jog enable pin, but I don’t think so.
04:15 PM cradek: thank you for working on finishing the ja merge...
04:17 PM andypugh: Well, I only went down this rabbit hole trying to figure out what the update_ini script can conclude from [TRAJ]COORDINATES and found myself in halui wondering what it was doing with it.
04:17 PM cradek: oh dear, that's a deep one
04:18 PM cradek: I'm on the road and am going to get to see seb_kuzminsky!
04:19 PM andypugh: Ahm you can hack this problem together, excellent :-)
04:20 PM cradek: I guess we never know what the future holds
04:20 PM cradek: back to driving, cheers
04:21 PM andypugh: Have fun