#linuxcnc-devel Logs

Dec 09 2022

01:00 AM pere: seb_kuzminsky: created <URL: https://bugs.debian.org/1025790 > to let the maintainer know.
02:19 AM -!- #linuxcnc-devel mode set to +v by ChanServ
02:50 AM -!- #linuxcnc-devel mode set to +v by ChanServ
04:27 AM -!- #linuxcnc-devel mode set to +v by ChanServ
09:24 AM -!- #linuxcnc-devel mode set to +v by ChanServ
09:25 AM seb_kuzminsky: pere: i didn't see the change documented anywhere, and actually the documentation is pretty ambiguous both before and after the change. So switching to explicitly checking the state (which *is* pretty well documented) seems better, like Craig said
09:27 AM pere: seb_kuzminsky: yeah.
09:28 AM seb_kuzminsky: thanks for reporting it - i think one of the things that makes debian awesome is that it improves communication between different open source projects like that
09:40 AM pere: it seem like a surprising change, so I just wanted to make sure the issue was publicly documented for all users to find, and that the maintainer knew about the risk in case it deserves a NEWS entry in the package.
09:40 AM pere: I doubt linuxcnc is the only program using this notation. :)
09:46 AM pere: seb_kuzminsky: any idea how to make the classicladder test more robust?
09:46 AM pere: or fix classiclasser on arm? I believe we need more tests.
09:49 AM seb_kuzminsky: i hope to spend some time debugging classicladder-on-arm next week
09:49 AM seb_kuzminsky: i agree more tests are always good
09:51 AM pere: it seem to be required to know that linuxcnc is working. :)
09:51 AM pere: I am very glad we discovered the ps bug early, before any users were affected.
09:51 AM pere: would have been great if we had a test for the arc issue, for example.
09:52 AM roguish[m]: what's a ps bug ?
09:52 AM pere: what is blocking people from writing tests? knowledge, skills, infrastructure, understanding or will?
09:53 AM pere: roguish[m]: <URL: https://github.com/LinuxCNC/linuxcnc/pull/2196 >
09:54 AM roguish[m]: ah, ok. just curious
09:55 AM pere: it was quite confusing to figure out. seb_kuzminsky found the cause yesterday, and at least me and him had fumbled in the dark for a few days.
09:55 AM pere: luckily we discovered the problem using the automatic testing of linuxcnc on github.
09:56 AM seb_kuzminsky: yeah, that testing infrastructure is working well and saving us from getting surprised when we prepare debian.org uploads
09:59 AM pere: I wish someone who understood classicladder would get involved in writing tests, to ensure most of its features are tested.
10:02 AM roguish[m]: seb_kuzminsky: good morning. a question please. the comp vfb_vfd.c started by padnos, finished by you. used for a Delta B type. I am using it for a Delta type E. the comms are the same and the modbus stuff used in the comp is the same. I would like to add some more modbus. is it possible? terribly difficult? if I start, can I ask you for spots of help?
10:09 AM seb_kuzminsky: pere: as far as i know there's not some silent expert (on classicladder or most anything else) hiding in the wings, waiting to be coaxed out of vacation. It's just a bunch of us halfwits muddling through as best we can. When something percolates to the top one of us tries to figure it out enough to push the project forwards a half step.
10:10 AM seb_kuzminsky: roguish[m]: if they use the same protocol it should be possible to extend the existing driver, yes
10:10 AM seb_kuzminsky: the challenge i think will be verifying that your changes work on both hardware devices. do you have one of each to test on?
10:12 AM roguish[m]: I only have my E. and if I have to create a new comp based on the B, that's ok. as i mentioned, the current B comp works for my E.
10:13 AM roguish[m]: I checked the docs as best I could before I tried it out. everything looked ok, and it must be as it does work.
10:13 AM seb_kuzminsky: roguish[m]: well that all sounds very promising!
10:15 AM seb_kuzminsky: one option would be to find a user that has a Delta VFD-B and is willing to test if your changes break their setup
10:15 AM pere: seb_kuzminsky: perhaps not vacating experts, but definitely busy experts not spending time writing tests. and I would love to help them find it easier to write tests.
10:16 AM seb_kuzminsky: another option is to fork the driver (ie copy vfdb_vfd.c to vfde_vfd.c) and make your changes in the new E driver, though this has the downside of twice as many drivers to maintain :-/
10:16 AM roguish[m]: any advice on modbus programming for idiots..?
10:16 AM roguish[m]: would be greatly appreciated.
10:16 AM roguish[m]: helps, or docs or books or anything?
10:16 AM seb_kuzminsky: pere: yeah, gotcha. imo it's currently pretty easy to write tests, but that may be blindness born from familiarity on my part
10:16 AM roguish[m]: if I actually get something done, I will post on the forum and ask for someone with a type B to check it out.....
10:17 AM roguish[m]: pere: seb_kuzminsky are the test written as scripts or full programs? got an example?
10:18 AM seb_kuzminsky: roguish[m]: there is general info about modbus on wikipedia: https://en.wikipedia.org/wiki/Modbus
10:18 AM seb_kuzminsky: linuxcnc uses libmodbus: https://libmodbus.org/
10:18 AM seb_kuzminsky: and as always, our existing code (and specifically the existing vfdb_vfd.c driver) can serve as examples
10:19 AM seb_kuzminsky: the tests can be written either as full programs, or as snippets of HAL configuration
10:20 AM seb_kuzminsky: our testing infrastucture is flexible and capable, but not well documented - i think the assumption has been that the existing tests act as examples, and that's been enough to get started writing new tests
10:21 AM seb_kuzminsky: i recognize that that is an unfair and unhelpful assumption
10:21 AM pere: roguish[m]: they can be scripts, hal fragments or full programs.
10:21 AM pere: roguish[m]: tests/ in the source are full of them, and scripts/runtest(.in) is the script running them.
10:21 AM roguish[m]: classic ladder is essentially a separate program, right? I used it long ago. and do understand ladder logic programming. not sure how one would set up tests.
10:22 AM pere: I have no idea if classicladder is a separate program, but got the impression that most of linuxcnc is separate programs.
10:22 AM seb_kuzminsky: classicladder is a separate HAL component, yes
10:22 AM pere: tests/classicladder/estop/ is the only test written so far, and it exposed a bug on arm alread. :)
10:23 AM seb_kuzminsky: it has a realtime part that you loadrt in hal, and an optional GUI that runs in non-realtime
10:24 AM seb_kuzminsky: bbl
11:19 AM JT-Shop_ is now known as JT-Shop
04:23 PM JT-Shop22 is now known as JT-Shop2
05:49 PM andypugh: Debian seems to be infested by obsessive pedants with an axe to grind. I don’t want to play any more.
05:50 PM andypugh: How much time did Adam Ant spend comparing files to find that one of our files looks quite a lot like another file from an even more obscure repository?
05:51 PM andypugh: He must have an axe to grind, and more time than us.
05:52 PM andypugh: We don’t need to be in Debian, it’s not worth the candle against this kind of attack.
05:52 PM andypugh: In my opinion.
05:53 PM andypugh: pere: Your thoughts?
05:56 PM CaptHindsight[m]: already in Gentoo :)
05:57 PM andypugh: Until twatface starts to attackus there too.
05:57 PM CaptHindsight[m]: not a problem
05:57 PM CaptHindsight[m]: lots of nuts everywhere
05:58 PM andypugh: Yeah, but if he can convince others that we don’t have legitimate copyright on our code then that’s it. We can’t prove that we do without going to court.
05:59 PM andypugh: (and I am not sure that we do anyway, it’s complicated)
06:01 PM CaptHindsight[m]: about 14 years ago I was asked to help out with a coil winder for ITW (Illinois Tool Works) near Chicago, the contractor i was going to replace found out that I had offered LCNC as a controller and be done in a few days, had a meltdown....
06:01 PM CaptHindsight[m]: he reported to wikipedia that LCNC was a dead project and should be removed
06:02 PM CaptHindsight[m]: they almost believed him
06:04 PM CaptHindsight[m]: LCNC has also had to put up with lots of hate from reprap, GRBL community crap as well
06:05 PM CaptHindsight[m]: I don't see a problem with 3rd party debs anyway
06:12 PM andypugh: Yeah, I am OK with us not being in Debian.
06:14 PM andypugh: And I suspect that this is just his opening salvo. He has already been on debian.legal expressing “concern” that we include Public Domain NIST code with our copyright on it. And that not all files contain a full License statement (hard to enforce for .comp files)
06:15 PM andypugh: All he has to do is sow doubt, and we can’t remove that doubt without recourse to a court case (which has still, afaik, never happened)
06:28 PM CaptHindsight[m]: he seems to be violating copyright as well with his nick :)
06:28 PM CaptHindsight[m]: Stuart Leslie Goddard might want a word woth him :)
06:29 PM CaptHindsight[m]: https://en.wikipedia.org/wiki/Adam_Ant
06:30 PM CaptHindsight[m]: https://en.wikipedia.org/wiki/Atom_Ant
06:33 PM andypugh: Yeah, unless he really is him. Which would be really disappointing as back in the day I was a real fan.
06:35 PM andypugh: https://youtu.be/Rm9drIwmmU4
07:02 PM Tom_L: one bad apple...
10:14 PM Guest1228 is now known as chrismorley
11:34 PM pere: andypugh: I am not as worried as you, as I know the enforcers of copyright in debian (the ftpmasters) to be reasonable people. I would make sure our documentation answer questions raised, for example by extending <URL: http://linuxcnc.org/docs/devel/html/common/emc-history.html >, which already have some key elements.
11:35 PM pere: I would also make sure we do a good job documenting correct licensing in our files, to make it easier to generate debian/copyright with correct information.
11:37 PM pere: the document mention the LinuxCNC Board of directors. Where is the list of the current members of this board? How are they selected?
11:39 PM pere: Is <URL: http://wiki.linuxcnc.org/cgi-bin/wiki.pl?BoardElection > the current board?