#linuxcnc-devel Logs

Jul 25 2018

12:58 AM CMorley: rmu: Qt_vcp_cleanup has been deleted. Enums work is not compatible with designer 5.5 which I use unfortunately.
01:00 AM CMorley: Feel free to submit a nicer screen - you are right the default screen is mostly used for testing
01:14 AM CMorley: spma: re toolwear offset editor - gladevcp's tooleditor handles tool wear in a bit better way - Gmoccapy uses it and I believe tooledit_widget.py can be run standalone using the python interpreter to load it.
01:23 AM hazzy-lab: shoot, missed chis, again
04:32 PM pcw_mesa: While being held hostage here (well actually waiting for his kids to be released from summer school) Amit Goradia has generously worked on
04:32 PM pcw_mesa: fixing some long standing mesa encoder bugs. I have his git patches, how can the be merged into 2.7,2.8?
04:49 PM seb_kuzminsky: pcw_mesa: the best way would be to make a fork of linuxcnc on github, push the changes to a branch there, and open a PR
04:52 PM seb_kuzminsky: does that make sense? i'm happy to try to describe any of those steps further
04:55 PM pcw_mesa: Yeah I understand generally...
04:56 PM pcw_mesa: do I need special permission to create a fork?
04:56 PM seb_kuzminsky: no special permission needed
04:56 PM pcw_mesa: (say mesa-encoder-bugfixes)
04:57 PM seb_kuzminsky: anyone can fork any public repo on github, without asking permission of anyone
04:58 PM pcw_mesa: OKIll see if Amit and I can figure out how to do that tomorrow when he's here
04:58 PM seb_kuzminsky: 'mesa-encoder-bugfixes' would be a good name for the branch in the forked repo
04:58 PM seb_kuzminsky: ok, i'll be online much of tomorrow if you have questions
04:59 PM pcw_mesa: Great, thanks!
07:37 PM * jepler_ finds that the hal session he started weeks ago to test the pid velocity estimate input is still running
07:37 PM jepler_: I'm so happy linuxcnc runs stably enough for that to be possible
07:37 PM jepler_ is now known as jepler
07:38 PM jepler: or maybe it wasn't weeks and I'm exaggerating, but still
09:05 PM skunkworks: i have had linuxcnc with axis up for weeks at a time. amazing
09:18 PM skunkworks: pcw_mesa has had systems up for years
10:04 PM linuxcnc-build_: build #3696 of 1405.rip-wheezy-armhf is complete: Failure [4failed compile] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1405.rip-wheezy-armhf/builds/3696 blamelist: Chris Morley <chrisinnanaimo@hotmail.com>
10:05 PM hazzy-lab: CMorley: Are you here?
10:05 PM CMorley: yes
10:05 PM hazzy-lab: hurray!
10:05 PM CMorley: :)
10:06 PM hazzy-lab: How are the STMBLs?
10:07 PM hazzy-lab: I got mine mostly assembled, but need to hunt down a right angle header for between LV and HV side
10:08 PM CMorley: I have one together and functional but setup has been difficult - I haven't put too much effort in though. documentation is lacking
10:08 PM CMorley: You were missing parts?
10:09 PM hazzy-lab: Just the little four pin headers, did Andy include them with your kits?
10:09 PM CMorley: yes
10:09 PM CMorley: he musta missed them
10:10 PM hazzy-lab: I should have some, just too lazy to hunt them down, lol
10:10 PM CMorley: lol
10:10 PM CMorley: do your motors use encoders or resolvers?
10:10 PM hazzy-lab: Resolvers
10:10 PM CMorley: mine too
10:11 PM hazzy-lab: I can't wait to get them running
10:11 PM CMorley: the phase setting seemed sensitive. if it wasn't close I got no feedback
10:11 PM CMorley: the default setting was .4 i changed it to .8 and boom feedack
10:11 PM CMorley: but getting closed loop has not happenned
10:12 PM CMorley: yes I'm excited too
10:12 PM hazzy-lab: That is good to know!
10:13 PM hazzy-lab: Hopefully you will be just ahead of me and figure out all the tricky parts :D
10:13 PM CMorley: ha - I'm waiting for you...
10:14 PM CMorley: It was great just seeing the motor move in open loop!
10:14 PM hazzy-lab: LOL
10:14 PM hazzy-lab: I bet!
10:17 PM CMorley: i have a blown servo drive on mt lathe i would like to replace with the STMBL - tired of using flacky old equipment
10:26 PM hazzy-lab: I am converting an old Mazak VTC-41, the drives work but its is going to be a pain to interface them with LCNC and they are so old I am worried the may not last much longer, so STMBL to the rescue!
10:26 PM linuxcnc-build_: build #5544 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/5544 blamelist: Chris Morley <chrisinnanaimo@hotmail.com>
10:31 PM CMorley: thats a nice size machine - you must have a large shop :)
10:35 PM hazzy-lab: alas, it is not my machine :(
10:37 PM hazzy-lab: My shop would almost fit inside the mazak, lol
10:40 PM CMorley: oh i see...well hopefully you get to use the converted machine :)
10:49 PM hazzy-lab: I will get to use it, if I ever finish it!
10:51 PM hazzy-lab: I have not had time to run qtvcp lately, but reading the commit messages it looks like you have been making a lot of progress!
10:53 PM CMorley: A lot of bug fixes on metric configs and rotary axis - stuff i hadn't tested
10:54 PM CMorley: I don't want to push it to master till it's to the point that I won't break it badly fixing stuff
10:55 PM CMorley: Someone on the forum tried it our and that swhen i got feedback about metric configs so that helps
10:58 PM CMorley: Know much about opengl programming? I tried to add gradient background to the graphics plotter but can't quite get it right
11:01 PM hazzy-lab: No I don't, but that would be very nice
11:04 PM hazzy-lab: Another thing that would modernize the look of the backplot would be antialiasing, looking at the QOpenGLWidget it looks like it is just setting a flag, but I tried and failed miserably
11:04 PM CMorley: yes it looks much better - i just can't get panning and rotation to work after
11:06 PM CMorley: ya i was just happy to get it to work at all in the beginning - now it would be nice to do better - mostly i needs to start with documenting how it works - since i don't lknow opwngl at all
11:06 PM CMorley: but that will have to wait for now
11:16 PM CMorley: Now I want to figureout how to make a widget I can double click and get a full screen or larger widget anyways. that and gestures for moving the view
11:17 PM hazzy-lab: Somebody complained about large files (~50mb) taking a long time to load in gremlin, so I did some tweaking to access vars directly instead of using method calls (Ex. self.is_lathe instead of self.is_lathe() ), and reportedly that sped it up by about 15 times
11:17 PM hazzy-lab: That is very significant, so I guess I should PR that
11:18 PM hazzy-lab: I was looking at the QGesture stuff yesterday, it is not very well documented :(
11:18 PM hazzy-lab: But for gremlin normal mouse events are probably sufficient
11:20 PM pcw_home: gremlin is the main source of pain and suffering on the RPI (200% CPU when running) Can I try your gremlin changes on the RPI?
11:21 PM CMorley: mouse movements are not good with a touch screen :)
11:22 PM CMorley: 15 times that is great
11:23 PM CMorley: I wonder if using cython would help - or coding more in c
11:26 PM hazzy-lab: pcw_home: rmu tried to test CMorley's Qt gremlin (or maybe it was mine??) on his Pi and could not get it to run. Something about the Pi using a different GL library
11:27 PM hazzy-lab: My changes are based on Chris's changes, so I am afraid it won't work
11:27 PM pcw_home: ahh this is QT gremlin
11:28 PM hazzy-lab: CMorley: I think Gremlin would be a prime place to use some Qt flavor C++!
11:28 PM CMorley: Sounds like the premise is the same though .. gremlin could get similar changes
11:28 PM CMorley: Ys I agree
11:29 PM hazzy-lab: CMorley: You are correct, but I think my changes only affect loading time, not the behavior one the file is loaded
11:32 PM CMorley: pcw-home : you have tested gmoccapy on the RPI then? The RPI can have disfferent distrbutions on it so maybe rmu didn't have the right one for gladevcp
11:33 PM CMorley: Kurt ahh I see
11:33 PM CMorley: gremlin actually uses c or c++ to convert the gcode to opengl calls - everything after that is python
11:34 PM pcw_home: I can try gmoccapy tomorrow
11:35 PM pcw_home: loading is slow on RPI also...
11:36 PM pcw_home: theres experimental opengl support on RPI (vs opengles I guess)
11:45 PM hazzy-lab: CMorley: Yes, the interpretor is c++, but that is blazing fast compared to the python code that generates the preview
11:45 PM CMorley: yes that is the problem with all the 'embedded' computers lack of opengl and graphocs drivers
11:46 PM CMorley: No there is a c++ file that converts the gcode to a display list and gremlin and sxis use that
11:46 PM pcw_home: I'll try enabling opengl on the RPI
11:47 PM CMorley: it's in the linuxcnc module IIRC
11:47 PM hazzy-lab: is that "rs274.glcanon.GLCanon"?
11:47 PM CMorley: pcw_home you are going to support the RPI or ar eyou just playing with it?
11:49 PM pcw_home: I have a couple RPI/RPI clone compatible daughtercards (one has been running linuxcnc/hm2_rpspi for a couple weeks now)
11:50 PM CMorley: hazzy-lab :emcmodule.cc
11:50 PM CMorley: hazzy-lab : logger methods
11:51 PM CMorley: pcw_home :cool - think there is a reasonable large market for them?
11:51 PM hazzy-lab: Aha, I assumed those were just for moving the tool!
11:51 PM pcw_home: main issue with RPI is Gremlin slowness/cpu hogging
11:52 PM pcw_home: I think so... kind of a Acorn like thing with a relatively fixed config