#linuxcnc-devel Logs

Sep 20 2018

08:15 AM Jin|away is now known as Jin^eLD
12:33 PM rene_dev_: andypugh how is multi spindle getting along?
01:10 PM andypugh: I haven’t really touched it since my email to the dev list
01:10 PM andypugh: Needs docs and shanges to runtests.
01:10 PM andypugh: (changes)
01:11 PM andypugh: And for a decision about whether 2.8 goes out soon, then I merge, or I merge now.
01:13 PM mozmck: andypugh: is it likely to break something else?
01:14 PM andypugh: Well, it shouldn’t.
01:14 PM mozmck: If not, I would say merge it now.
01:14 PM andypugh: It fails a lot of “runtests” now, but that is because of an an added index to the spindle commands.
01:16 PM andypugh: I am busy tonight, but will look to get it merged by sunday evening.
01:16 PM mozmck: I see. is that index required?
01:17 PM mozmck: Or are the tests not expecting it and failing because it is there?
01:17 PM andypugh: Yes, it indicates in the commands which spindle is being used
01:17 PM andypugh: Yes, that.
01:17 PM mozmck: Ok, so if you only have one spindle you don't need the spindle index then.
01:19 PM andypugh: The falures are purely with the comparison between actual and predicted canon logs.
01:20 PM andypugh: Oh, I also need to modify the update-ini script and run the sample configs through it.
01:22 PM hazzy-m: andypugh: Does it add/change any of the python interface attributes? (I would assume some things need to be changed to tuples etc. to support values from multiple spindles)
01:22 PM andypugh: I confess that I have forgotten.
01:22 PM andypugh: I would expect so.
01:23 PM hazzy-m: ok 🙂
01:23 PM mozmck: I kind of hate to push off state-tags, external-offsets, and reverse-run, but perhaps we can release and then merge those right away and work on a new release with those features.
01:23 PM mozmck: (as well as others I didn't mention...)
01:28 PM rene_dev_: none of those features should break anything, so might as well be merged ;D
01:28 PM rene_dev_: unless they are used, that is.
01:28 PM mozmck: ya think? ;-)
01:29 PM hazzy-m: state-tags being a possible exception to that
01:29 PM mozmck: That's like saying, "that food is non-fattening if you don't eat it"
01:29 PM rene_dev_: who is in charge of releasing 2.8?
01:29 PM rene_dev_: yeah^^
01:29 PM mozmck: technically myself, but I have been very busy and slow about doing anything about it.
01:30 PM rene_dev_: Honestly I dont really care, as I use master anyway. but would be nice to have, for normal users
01:30 PM rene_dev_: none of my machines work with 2.7
01:30 PM hazzy-m: I suspect releasing 2.8 will bring some life and interest back to the project, its been kinda stagnant for a while, without any significant improvements being released
01:31 PM mozmck: Probably true.
01:31 PM rene_dev_: agree
01:31 PM rene_dev_: and it is really stable as it is right now
01:32 PM hazzy-m: I think it is more stable than 2.7, lol
01:32 PM mozmck: So I think if we merge andy's spindle stuff, and any other bug fixes like you suggested rene-dev, then we should release as soon as possible now.
01:32 PM pcw_mesa: Ive had 2.8 and Preempt-RT up for months with Axis
01:33 PM pcw_mesa: not running a wide range of gcode however
01:33 PM hazzy-m: and when we do release 2.8 we will have a dev branch again 😃
01:33 PM pcw_mesa: and > 1 month on a RPI3
01:35 PM mozmck: So far I have not had time to actually run 2.8 on a machine myself, but 2.7 works well for me :-)
01:35 PM andypugh: <shuffles feet>
01:36 PM andypugh: I might have forgotten to convert the Python interface to multispindle
01:36 PM * hazzy-m has never run 2.7 on a real machine
01:36 PM andypugh: But then the Python interface isn’t very good for spindles. I can’t actually see how you set a spindle speed….
01:37 PM andypugh: I think you just have to send SPINDLE_INCREASE rpm / 100 times...
01:37 PM hazzy-m: You can't :(
01:37 PM hazzy-m: it is even hard to get the commanded spindle speed
01:38 PM hazzy-m: have to do stat.settings[1] * stat.spindlerate
01:38 PM andypugh: c.MDI(“M3 S1000 $2”) ought to work
01:38 PM hazzy-m: which is not intuitive
01:39 PM hazzy-m: sure, but there are commands to turn the spindle on and off in manual mode
01:39 PM andypugh: Yes, I need to add the Python interface to the list. Looks like I have a busy Sunday ahead :-)
01:40 PM hazzy-m: you can do cmd.spindle(linuxcnc.SPINDLE_FORWARD, speed)
01:40 PM hazzy-m: (I just added that to the docs, took me for ever to figure that out, lol)
01:41 PM hazzy-m: So I would assume with multi spindle you would so something like spindle(spindle_num, linuxcnc.SPINDLE_FORWARD, speed)
01:42 PM mozmck: forever? it was about 3 minutes 33 seconds between your "You can't" and telling us how!
01:43 PM mozmck: 2:33
01:43 PM mozmck: :-)
01:49 PM andypugh: Yes, looks like a simple fix in emcmodule.cc
01:49 PM andypugh: (the other end already knows what to do)
01:50 PM hazzy-m: mozmck: What I meant is you can't set the requested spindle speed without actually starting the spindle, (no equivalent to the S word). The `speed` argument is optional in cmd.spindle, so it would make since if there was a way to set it, in stead it seems to always default to 1rpm
01:52 PM hazzy-m: It took me about 12 hr to even figure out that I could pass a speed argument to cmd.spindle, as it was not mentioned in the docs anywhere
01:53 PM hazzy-m: probably the best thing to do there is make the speed parameter required ...
01:55 PM seb_kuzminsky: i'm not a very active developer in the linuxcnc world these days, and I sure appreciate you all picking up the gauntlet
01:56 PM seb_kuzminsky: i will be happy to help with release mechanics, git stuff, and buildbot stuff, in a supporting role, while y'all take the lead
01:56 PM seb_kuzminsky: i specifically want to comment on hazzy-m's statement that "when we do release 2.8 we will have a dev branch again":
01:57 PM seb_kuzminsky: i recommend making the 'stable/2.8' branch or 'stable/3' or whatever you want to call it early in the release process, as it signals the user base that the stable release is nearing, and encourages everyone to increase the dilligence of our collective testing and bug reporting
01:58 PM seb_kuzminsky: as soon as the stable release branch is created, master becomes the "dev" branch again
01:59 PM seb_kuzminsky: *however* i know that lots of folks run master rather than the stable branches, so I encourage you to not break master; do scary development work in feature branches and merge them into master when they're ready
02:05 PM rene_dev_: hazzy-m what UI stuff are you working on at the moment?
02:07 PM mozmck: seb_kuzminsky: yeah, good idea. I think it is kinda working like that anyhow with people using feature branches.
02:08 PM rene_dev_: what then usually happens, is that once it works it stays there forever....
02:08 PM hazzy-m: seb_kuzminsky: I completely concur with that. I think we should feel a little more relaxed about merging not quite polished features into master than we have been (to expose them to a wider range of testers), but also strive to keep master as reliable as possible. After all, this is a machine control, and bad things can happen
02:08 PM seb_kuzminsky: mozmck: i agree, i just want to re-state my support for that development model
02:08 PM seb_kuzminsky: hazzy-m: +1
02:09 PM mozmck: +2
02:09 PM hazzy-m: rene_dev_: QtPyVCP, a branch of CMorley's QtVCP: https://gitlab.com/Hazzy/qtpyvcp
02:09 PM rene_dev_: so you will make a new hazzy version based on that?
02:10 PM rene_dev_: how is that getting along? especially the backplot...
02:10 PM seb_kuzminsky: there was been a general lack of energy/attention from the "old boys club" (jepler, cradek, and me, pretty much), so stuff has been languishing in feature branches for a long time, from lack of volunteer time to test and integrate stuff
02:11 PM seb_kuzminsky: so again, i'm pleased and hopeful that a new generation of energetic & motivated developers will step up to carry the torch for a while as us old-timers drift away
02:13 PM hazzy-m: rene_dev_: Yes, hazzy3 🙂 It is coming very well, home to be able to run a machine with it very soon. I haven't done any more on the backplot, but plan to once some of the other things are out of the way
02:13 PM rene_dev_: cool :)
02:14 PM rene_dev_: seb_kuzminsky yeah, I know what its like when you work on a project for too long :D
02:14 PM rene_dev_: do you know who keeps track of the pathpilot changes, and puts them back into linuxcnc? do you know if they have a public git repo of their changes?
02:14 PM seb_kuzminsky: for me, it's more that i want to get deeper into *using* linuxcnc for other projects, rather than having linuxcnc *be* my project all the time
02:15 PM rene_dev_: haha, I feel the same with stmbl sometimes :D
02:15 PM seb_kuzminsky: rene_dev_: pathpilot does not have a public git repo, as of last time i checked, but they're legally and morally obligated to provide the source code to their fork of linuxcnc when asked
02:15 PM hazzy-m: rene_dev_: Lcvette is working on a pathpilotish interface built with QtPyVCP: https://forum.linuxcnc.org/41-guis/34996-another-gui-project-underway?start=70
02:16 PM seb_kuzminsky: so you can email them and they *should* send you a CD with their code, possibly charging you for shipping
02:16 PM rene_dev_: none of my machines have a real clean cabinet, with a stmbl with case
02:16 PM rene_dev_: not like this: https://imgur.com/a/19afQ9S
02:16 PM seb_kuzminsky: i think zultron and rob ellenberg have occasionally tried to offer PRs of some of pathpilot's changes, and received little or no feedback from the linuxcnc dev community
02:17 PM rene_dev_: Im not even talking about the gui, but from the linuxcnc. apparently they fixed some things.
02:17 PM rene_dev_: and their stuff should run stable...
05:39 PM Jin^eLD is now known as Jin|away
08:43 PM memfrob: Hi all, scripts/rtapi.conf.in says "DO NOT EDIT THIS FILE" but RTAI 4 is missing from there
08:44 PM memfrob: I believe I tried this though and it didn't work: http://dpaste.com/1MA0P6H
08:45 PM memfrob: It's a very fragile file and was looking for suggestions.
09:15 PM jepler: looks like the comment in that file is a lie
09:51 PM memfrob: I'll ignore it then, thanks! Also it seems that || doesn't work with that.
09:51 PM memfrob: 6+.0+6
11:33 PM memfrob: take care all, night.