#linuxcnc-devel Logs
Sep 12 2022
#linuxcnc-devel Calendar
03:50 AM pere: hi
08:56 AM pere: has there been any work to see if there are new machinekit components that would be useful to add to linuxcnc?
11:58 AM pere: I notice there are 37 components in machinekit that are not in linuxcnc, and 104 'v2' components that might be in linuxcnc already.
12:17 PM silopolis[m]: Hi
12:17 PM silopolis[m]: pere: interesting...
12:22 PM silopolis[m]: Only knew about HAL/emc-apps split, cmake and haltalk/machinetalk
12:23 PM silopolis[m]: Would be interesting to have a detailed comparison table if it doesn't exist
12:30 PM pere: I also notice machinekit hal support 64 bit types.
12:39 PM pere: and something called pin_ptr
12:48 PM pere: and 42 components in linuxcnc that are not in machinekit.
12:48 PM pere: note, I do not handle renames, so dbounce and debounce are considered different components. not sure if that is accurate.
12:51 PM pere: <URL: https://github.com/LinuxCNC/linuxcnc/pull/2030 > is an example.
12:51 PM silopolis[m]: Didn't got much feedback for my reformatting PR from the "docs team" !? 🤔
12:52 PM pere: silopolis[m]: too many PRs, almost everyone I look at got a merge conflict...
12:52 PM pere: no idea which one you talk about.
12:52 PM pere: I suspect all these 'reformat' PRs is a diversion until we can formulate some style guide.
12:53 PM Tom_L: pere, no luck with lb. maybe pi-gen can be modified for arm64
12:53 PM silopolis[m]: https://github.com/LinuxCNC/linuxcnc/pull/1742
12:53 PM Tom_L: afraid it' above my pay grade though
12:54 PM pere: I suspect I need to find a machine with a lot more space space to have a look myself.
12:55 PM Tom_L: i think the bootloader sequence is too different for the lb config
12:55 PM Tom_L: take a peek at linuxcnc/pi-gen if you want
12:56 PM silopolis[m]: pere: I believe they are part of style guide development
12:56 PM pere: silopolis[m]: I notice hans asked if it was ready to be merged. did he git a clear answer?
12:57 PM pere: silopolis[m]: well, I believe 'style guide development' need to result in written down descriptions in a adoc file in git...
01:03 PM silopolis[m]: <pere> "silopolis: I notice hans asked..." <- Sent a message end of last week and switched PR from draft to ready
01:04 PM pere: silopolis[m]: ack. now it got a git merge conflict.
01:05 PM silopolis[m]: <pere> "silopolis: well, I believe '..." <- Of course, but testing and agreeing on ways to do and improve formating is best done through actual work
01:05 PM pere: sure. I am just missing the last step, getting it into a adoc file in master.
01:06 PM silopolis[m]: pere: Seen, rebased this morning but have to do it again
01:06 PM silopolis[m]: pere: Will work on this next
01:09 PM silopolis[m]: But if I could get the remaining work triggered by this PR before and avoid more painful rebases... I'd really be grateful
01:20 PM pere: I am not competent to evaluate that pull request myself, so hope hans can do it.
01:54 PM pere: It would have been great if the entire documentation team was here.
03:30 PM -!- #linuxcnc-devel mode set to +v by ChanServ
04:47 PM JT-Shop22 is now known as JT-Shop2
06:01 PM skunkworks: andypugh: pretty sure CaptHindsight[m] had the same crash spinning the encoder on rtai..
06:02 PM skunkworks: I don't have an rtai build here to test.
06:03 PM andypugh: Yes, but I don’t have any better ideas what to do to reproduce the problem
06:04 PM skunkworks: I suppose you don't have a parallel port and an encoder? ;)
06:05 PM skunkworks: will your rtai install run a 10us base thread? I could experiment with a longer thread?
06:16 PM skunkworks: andypugh: anything I could do? Could I some how break the code when it happens? (or is that too hard to walk though?)
06:18 PM andypugh: I am getting frustrated, as I can’t stop LinuxCNC saying “RTAPI requires the real-time kernel 4.19.195-rtai-amd64 to run. Before running
06:18 PM andypugh: this realtime application, reboot and choose this kernel at the boot menu."
06:20 PM andypugh: Oh, it was the rip-environment script… I didn’t re-run it after the reboot.
06:21 PM andypugh: And it crashed immediately.
06:21 PM andypugh: Good
06:21 PM skunkworks: Oh - good!
06:23 PM skunkworks: I am in the process of setting up a mouse sensor on the emco encoder wheel to test the missing tooth encoder setup.
06:27 PM skunkworks: but it will be a few days
06:30 PM mrec: LBP16_INIT_PACKET4(packet, 0x4983, 0x0002); does anyone know where those commands are handled in the firmware?
06:30 PM mrec: mesa firmware
06:36 PM CaptHindsight[m]: skunkworks: I haven't tested RTAI on 2.9 yet and Bullseye with the encode and LPT. RTAI on 2.8 Buster did NOT generate the signal 8 core dump
06:38 PM CaptHindsight[m]: I was able to generate the signal 8 core dump on preempt_rt with 2.9 on Bullseye at times by just plugging the LPT cable into the connector on the card, so just about any signal will do it
06:38 PM CaptHindsight[m]: mrec: pcw is the one to ask
06:42 PM skunkworks: CaptHindsight[m]: I think andypugh got it to fail.
06:43 PM andypugh: Yeah, it’s a simple divide-by-zero
06:43 PM CaptHindsight[m]: skunkworks: sometimes it happens before I even enable the machine, might be the encoder wheel just vibrating
06:44 PM andypugh: Yes, possibly a reversal where there is an edge but no counts change.
06:44 PM andypugh: ie, edges have been seen, but counts have not changed.
06:44 PM CaptHindsight[m]: Alec did all the 2.9 RTAI his PC, I haven't installed it on anything yet
06:45 PM CaptHindsight[m]: I do have to go back and test the CH382 with 2.8 RTAI since I did find one bad ch382 board here
06:47 PM CaptHindsight[m]: I'd see pulses generated in halscope but nothing to the stepper drive
06:47 PM CaptHindsight[m]: so it has bad outputs, the encoder inputs came through to the GUI just fine
06:48 PM CaptHindsight[m]: http://wiki.linuxcnc.org/cgi-bin/wiki.pl?Parallel_Port_Cards I'm keeping track of my LPT card tests here
06:48 PM CaptHindsight[m]: feel free to add
07:07 PM CaptHindsight[m]: i just tested a known good ch382 with RTAI on 2.8 Buster and the encoder signals make it into halscope but not the spindle RPM gui in Axis
07:36 PM skunkworks: andypugh: thank you!
07:53 PM skunkworks: CaptHindsight[m]: what? weird.
07:54 PM CaptHindsight[m]: skunkworks: going to install RTAI on Bullseye later with 2.9
07:56 PM CaptHindsight[m]: not too concerned about it until after the encoder bug gets fixed
07:57 PM skunkworks: CaptHindsight[m]: so far so good. no error