#linuxcnc-devel Logs

Oct 15 2019

03:59 PM TurBoss: hi!
04:07 PM Lcvette: need some devel help in linuxcnc for an issue popping up with work offsets and the status changing to G54 at the start of a program even though the current status should be the same as the currently active work offset, it is making it quite difficult to have the offsets work in the new gui we have developed. I would exdpect that the M2 command that reverts the G5x offset back to G54 at program end would not activate
04:07 PM Lcvette: a status change until that M2 command was executed, but it seems it is executing at cycle start which does not seem correct. can anyone here help with that?
04:08 PM Lcvette: is anybody here?
04:11 PM andypugh: Hmm, it may be that you are falling foul of the two different parts of LinuxCNC not always appearing to be in synch
04:11 PM Lcvette: not sure i understand falling foul
04:12 PM Lcvette: its wreaking havoc on the gui display
04:12 PM andypugh: The interpreter tends to run straight through the G-code, sending all the moves to the queue. The machine will do things in order, but the status you see in the GUI might be the exit status of the interpreter rather than the active status of the machine.
04:12 PM Lcvette: and work offsets
04:12 PM Lcvette: backplot
04:13 PM andypugh: (I might be completely wrong here, of course)
04:14 PM Lcvette: thats what i assumed was happening
04:14 PM andypugh: Do you see the same thing if there is a toolchange before the M2?
04:14 PM Lcvette: but then how can i use a status channel for keeping the gui correct?
04:15 PM Lcvette: ie at program end when the offsets change on the machine, that signal is not available?
04:16 PM andypugh: I don’t know. I can only suggest looking to see if you an work out what other GUIs, such as Axis, do.
04:16 PM Lcvette: the only time i have been successful so far is by removing m2/m30 and using %gcode%
04:16 PM Lcvette: i checked and axis is the same as our gui in respect to the synch issue
04:16 PM Lcvette: it shows active gcodes as G54 even while running a G55 program
04:16 PM andypugh: (And % has its own problems, unless that has been fixed, with subroutines)
04:17 PM Lcvette: is there another signal that can be tapped in linuxcnc to appropriately update the gui?
04:18 PM linuxcnc-build: build #6221 of 1201.rip-lucid-rtai-i386 is complete: Failure [4failed compile] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1201.rip-lucid-rtai-i386/builds/6221 blamelist: andypugh <andy@bodgesoc.org>
04:19 PM Lcvette: its a bit disconcerting to see G54 offset illuminate while a G55 program is underway
04:19 PM andypugh: Chris Morley might know, he’s been ding a lot og GUI work recently.
04:20 PM Lcvette: andypugh: thanks!
04:20 PM Lcvette: Hopefully he pops in and can shed some light
04:20 PM andypugh: He is seen here occasionally, but it possibly best contacted on the forum or the mailing list.
04:21 PM Lcvette: Ok thanks for that!
04:21 PM linuxcnc-build: build #5427 of 1301.rip-precise-rtai-i386 is complete: Failure [4failed compile] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1301.rip-precise-rtai-i386/builds/5427 blamelist: andypugh <andy@bodgesoc.org>
04:21 PM Lcvette: By chance do you know if the offset change command in m2 can be removed?
04:22 PM Lcvette: Or m2 customized?
04:23 PM Lcvette: Or linuxcnc set to accept a custom user mcode instead as program end?
04:24 PM andypugh: I think it is expected by the standard.
04:24 PM Tom_L: yeah but he doesn't like the standard :)
04:24 PM Lcvette: it is, we found that in the NIST
04:24 PM andypugh: M2 is not in the list of remappable codes. http://linuxcnc.org/docs/2.7/html/remap/remap.html#remap:remappable-codes
04:25 PM Lcvette: it would be fine if it was synched to the machine actions
04:25 PM linuxcnc-build: build #4040 of 1401.rip-wheezy-rtai-i386 is complete: Failure [4failed compile] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1401.rip-wheezy-rtai-i386/builds/4040 blamelist: andypugh <andy@bodgesoc.org>
04:25 PM Lcvette: but being the interperetor runs through and changes the active gcode at cycle start it is a bit problematic
04:27 PM andypugh: It probably doesn’t in G-code with tool changes or other queue-busters in it.
04:27 PM andypugh: (That’s speculation, I would be interested to see if it is true)
04:27 PM Lcvette: i will test that, but i would thing the moment the last que buster was implemented it would again revert to G54 in status
04:28 PM Lcvette: as im sure the interperter would run through the remaining code and find m2 again
04:28 PM Lcvette: so it would just be delaying the inevitable and the same gui display issue would occur
04:28 PM Lcvette: at least in speculation
04:30 PM andypugh: Yes, that is what I would expect, if the problem is what we are speculating that it might be
04:31 PM Lcvette: hopefully ChrisM has a work around
04:31 PM Lcvette: this is driving me nuts
04:31 PM Lcvette: lol
04:31 PM linuxcnc-build: build #3360 of 1902.clang-wheezy-rtai-i386 is complete: Failure [4failed compile] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1902.clang-wheezy-rtai-i386/builds/3360 blamelist: andypugh <andy@bodgesoc.org>
04:41 PM linuxcnc-build: build #6225 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/6225 blamelist: andypugh <andy@bodgesoc.org>
05:25 PM CMorley: If you are talking of the fact status reports the task status rather then the current status, there is no work around. There is a branch to address this problem called state tags. My thought would be to merge it to master and work on it - after 2.8 is released.
06:10 PM Lcvette: That's exactly what I'm referring too!
06:13 PM Lcvette: Lol
06:51 PM Lcvette: That is exactly what I am referring too
06:59 PM Lcvette: CMorley: how are you connecting the GUI to the any offset buttons so they stay correct and current?
07:03 PM CMorley: I use hal_glib - which just polls linuxcnc's status - there is no other way I know that you can. Most programs I test with don't chamge the origin system mid program so I didn't notice the problem
07:04 PM CMorley: So in shorter words - they don't stay correct
07:04 PM Tom_L: he's finding that out :)
07:05 PM jthornton: sounds like state tags is a good thing to have
07:05 PM CMorley: that's what statetags are used to fix. Linuxcnc is behind the times
07:05 PM Lcvette: We can pull the status out using the python interface status:gcodes?string and each button has a checked if 'G5n' in ch [0]
07:06 PM Lcvette: Ah
07:07 PM CMorley: I thing you would find pilotpath uses statetags - I'm sure that's where the code came from
07:08 PM Lcvette: Yeah we were noticing because we previously only had mdi and gcode change the offsets from linuxcnc only and when we stayed running programs, the backplot got whacky
07:08 PM Lcvette: https://youtu.be/qdY2Lzsq5Cg
07:11 PM Lcvette: It isn't an issue if you remove the program ending M2 or M30
07:11 PM CMorley: On one of qtvcp's test screens it display the task state - reading,waiting etc. With a long program you can see exactly the effect - when task is finished reading the gcode status changes (to end of program states)
07:13 PM Lcvette: Interesting idea
07:13 PM CMorley: Anyways there is no practical work around. We need more linuxcnc developers so we can take on these features.
07:14 PM Lcvette: Maybe when I bought learn code
07:14 PM Lcvette: Maybe when I finish learning code
07:15 PM Lcvette: I'm illiterate at code
07:15 PM Lcvette: Lol
07:15 PM Lcvette: I'm just the picture jockey
07:16 PM CMorley: just find a little corner to polish and work from there - i'm a self learned programmer - I'm not great but have learned alot along the way.
07:16 PM TurBoss: state-tags-master to master merge looks scary
07:17 PM CMorley: I bet it will be - thats what happens when you wait 5 years to merge code :)
07:17 PM Lcvette: :o
07:17 PM Lcvette: Donut turboss
07:17 PM Lcvette: Do it turboss
07:18 PM TurBoss: it will take a bit...
07:18 PM Lcvette: It's needs hero
07:19 PM Lcvette: CMorley: thanks for the answer and letting me know I'm not crazy
07:19 PM CMorley: I personally feel linuxcnc as a project is in a precarious state - not enough developers
07:19 PM CMorley: Your not crazy lol
07:19 PM CMorley: I've been hoping for staetags for years too.
07:19 PM CMorley: have a good day
07:20 PM Lcvette: I hope these new guis will bring in some new popularity and another generation of developers
07:20 PM Lcvette: Thanks again
07:48 PM skunkworks: there is some conversations on github about state tags
07:48 PM Lcvette: yeah?
07:48 PM skunkworks: rob and john would be the ones to ask
07:48 PM Lcvette: ok cool
07:49 PM Lcvette: turboss ^^
07:49 PM TurBoss: i saw
07:49 PM skunkworks: https://github.com/LinuxCNC/linuxcnc/issues/134
07:50 PM TurBoss: ty
10:19 PM TurBoss: I manage to rebase and compile state_tags branch on latest master
10:28 PM Tom_L: that was quick