#linuxcnc-devel Logs

Nov 26 2022

11:11 AM seb_kuzminsky: `checkin` is a new kind of buildbot thing called a LocalWorker, i wonder why it keeps going missing...
11:28 AM pere: wow, my reminder to the emc-developer@ mailing list about the video chat on monday has been rejected with this comment: "Post by non-member to a members-only list"
11:50 AM roguish[m]: pere: become a member....
11:52 AM pere: Perhaps <URL: https://bugs.debian.org/1001904 > need to be reopened...
12:20 PM roguish[m]: pere: https://sourceforge.net/projects/emc/lists/emc-developers
12:21 PM roguish[m]: https://sourceforge.net/p/emc/mailman/
12:28 PM pere: roguish[m]: is this related to the debian bug?
12:28 PM roguish[m]: no, the developers mail list....
12:34 PM pere: the debian bug report is also related to the developers mailing list.
12:34 PM pere: if non-spam emails from non-subscribers now are rejected, the contact point of several debian packages need to change to a working mail address.
12:36 PM roguish[m]: don't know about that.... but you should consider officially joining. I've been getting the emails for users and developers for like 15 years now.....
12:44 PM pere: the bug report got all the links required to learn more.
01:16 PM seb_kuzminsky: pere: i rejected it as a favor to you, to remind you to sign up ;-)
01:19 PM seb_kuzminsky: i've been letting emails from debian robots through, and emails from non-linuxcnc folks responding to threads from them
01:22 PM pere: haha. and I posted as a favor to the project, but do not really care much if everyone remember the meeting monday.
01:25 PM seb_kuzminsky: pere: help me understand - are you opposed to signing up for the mailing list? do you want me to pass your emails through from now on?
01:26 PM seb_kuzminsky: i stop a couple of spam emails per day from going to that list, and i know some other folks do too, but this is a job i'd rather not have
01:31 PM pere: seb_kuzminsky: not really opposed, but did not bother as long as I was unable to reach linuxcnc.org because of its access filtering.
01:31 PM pere: If the web site did not allow me to access it, I did not bother using it or its mailing list.
01:32 PM seb_kuzminsky: the mailing list and the website are hosted by different providers, are they both filtering tor?
01:33 PM pere: the last three times I have by used linuxcnc.org it has worked, so perhaps the issue has been fixed? Do not really follow the mailing list, did not test its access.
01:35 PM seb_kuzminsky: the Dreamhost setup for linuxcnc.org is a mystery to me and i haven't talked to the hosting provider about anything, ever. I'm glad to hear it magically fixed itself for you.
01:36 PM seb_kuzminsky: soo... do you want to stay unsubscribed to the mailing list, and have me approve each of your emails by hand?
01:36 PM pere: I do not know if it will last, as I was told earlier that the web site worked for others via Tor while I was blocked, so I guess the issue might reappear any time unless really fixed in the configuration somewhere.
01:38 PM seb_kuzminsky: i don't think the access trouble you're (intermittently?) having with wlo is anything that i can help with
01:38 PM seb_kuzminsky: i do want to help you with the mailing list though
01:38 PM pere: I am perfectly happy with limiting my contributions to IRC and the github repo, and do not really need to use the mailing list. but I was asked to organize the next video chat, and it seemed like a nice thing to do to tell the mailing list subscribers too.
01:40 PM pere: I am sure I could get on the mailing list, but just decided fairly randomly that if the project did not want Tor users to use their web pages, then I dropped subscribing to the mailing list.
01:40 PM pere: And I get way to much email already, so it was not really a big loss to me to receive less email than I could. :)
01:41 PM pere: I am sure I fail to get a lot of politics and discussions, which I am perfectly happy to skip as well. :)
01:42 PM pere: As the web site seem to be working now, I am considering joining the list. perhaps after monday. :)
01:43 PM seb_kuzminsky: ok, sounds good
02:15 PM roguish[m]: pere: just added the video meeting to my calendar. i'll watch and listen, unless it gets crowded. If it's crowded I'll drop off to save bandwidth. 😎
03:01 PM linuxcnc-build2: Worker `checkin` is missing. It was seen last on Sat Nov 26 12:59:39 2022.
03:02 PM pere: rene-dev5: around?
03:03 PM pere: I've been unable to figure out how to write a tests/ script for the classicladder, ref <URL: https://github.com/LinuxCNC/linuxcnc/pull/1657 >. Anyone here able to help me?
03:25 PM seb_kuzminsky: pere: thanks for working on tests for the new classicladder
03:26 PM seb_kuzminsky: if i were doing it, i'd focus on the good old "provide a specific input pattern, watch the output and make sure it behaves as expected"
03:26 PM pere: the core of the problem is that I do not understand classicladder at all, so not sure how to test 'behaves as expected'.
03:27 PM seb_kuzminsky: yeah, understood. i don't know much about classicladder either. cradek pointed me at it years ago and i never followed up on it much... You'll learn a lot from doing this testing, then you can educate me :-)
03:28 PM seb_kuzminsky: for the test pattern i'm talking about, see e.g. tests/multiclick
03:33 PM pere: looking
03:34 PM pere: My test approach is to run linuxcnc with a python script as display, and use it to poke the pins and see if hal get the expected signals. but I do not know what to do...
03:36 PM seb_kuzminsky: yeah, a python "gui" that pretends to be a human peeking & poking at linuxcnc is a good way to implement a test
03:36 PM seb_kuzminsky: tests/motion has a bunch of tests like that
03:37 PM pere: I believe I got the test framework fairly under control, and can set up a simple ini and hal file, but no idea how to connect a classicladder setup into the hal pins and know what to expect...
03:37 PM seb_kuzminsky: have you seen the ladder docs? e.g. http://linuxcnc.org/docs/devel/html/ladder/ladder-intro.html
03:41 PM seb_kuzminsky: i see that there are no(!) tests involving classicladder yet, but if you look at configs/sim/axis/classicladder there's an example of how to hook it up there
03:43 PM pere: I found the example, tried to replicate it in a test tonight, but still some typos left. :)
03:43 PM pere: I have not read the ladder docs yet.
03:46 PM andypugh: I have a user saying that LinuxCNC does not make circles for him since the large-arc fix.
03:47 PM pere: calculation error?
03:47 PM andypugh: After homing , the machine stay on x0y0z0.
03:47 PM andypugh: Comand in mdi g2 x0y0z-1j4 f1000, the machine mocve only in z-1
03:47 PM andypugh: command now g0 z0
03:47 PM andypugh: command again g2 x0y0z-1j4 f1000 and now the circle strt z 0 finisc
03:47 PM andypugh: complete at z-1.
03:47 PM pere: andypugh: something to write a test in tests/ for?
03:48 PM andypugh: The thing is, I absolutely can not reproduce the problem. Here are two videos of me not reproducing the problem
03:48 PM andypugh: https://youtu.be/sOs4lATKm7Y
03:48 PM andypugh: https://youtube.com/shorts/DdJKSPb4KlA
03:49 PM seb_kuzminsky: what are the user's machine's X & Y limits? maybe the circle doesn't fit in their work envelope?
03:49 PM andypugh: it’s an 8mm circle
03:49 PM andypugh: And it isn’t that it errors, it just makes a Z move with no circle.
03:50 PM seb_kuzminsky: i remember cradek spent a lot of time making circles work when the startpoint & endpoint were super close
03:50 PM andypugh: Yeah, then I broke it, and fixed it.
03:51 PM andypugh: Because we had a problem with relatively short, large radius curves.
03:51 PM Tom_L: if the circle was outside the machine limits it would error on loading the file
03:52 PM andypugh: Does anyone have a live LinuxCNC to play with?
03:52 PM andypugh: It’s easy to test, home the machine and do a g2 x0y0z-1j4 f1000
03:53 PM Tom_L: metric?
03:53 PM andypugh: If it makes a spiral, it’s good. If it only moves in Z, then it’s the reported bug.
03:53 PM Tom_L: mine won't do 1000ipm
03:53 PM andypugh: Yeah, metric
03:53 PM andypugh: Should be OK in a sim.
03:54 PM seb_kuzminsky: andypugh: which branch is the user on?
03:57 PM Tom_L: pretty sure my test pc is 2.9 pre
03:59 PM andypugh: He was on master/2.9.0-pre0 and I was wondering if he was somehow missing https://github.com/LinuxCNC/linuxcnc/commit/23bb2ec49c1d75296b88dcae91a6fd95abdcd1b0 bit he is now on 2.10, latest master
04:00 PM Tom_L: mine went straight down in Z
04:01 PM Tom_L: 2nd test made an arc
04:01 PM Tom_L: (helix)
04:02 PM Tom_L: after i gave it an G0 x0y0z0
04:03 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/arc_test.png
04:03 PM seb_kuzminsky: same here
04:04 PM seb_kuzminsky: i'm on the tip of 2.9, sim/axis/axis. First G2 went straight down, g0 back up to z0, second G2 made the expected helix
04:04 PM Tom_L: sounds the same
04:05 PM Tom_L: i did the png for the commands and result
04:05 PM Tom_L: and exact lcnc version
04:06 PM Tom_L: pic always worth 1k words :)
04:06 PM andypugh: OK, so you chaps can reproduce the problem. Now why can’t I?
04:06 PM Tom_L: are you sending it to zero first?
04:07 PM andypugh: No, just homing. I have reconfigured the sim/axis_mm so that it homes to 0,0,0 without any motion.
04:08 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/arc_test1.png
04:08 PM Tom_L: with just the g20 & arc command
04:08 PM Tom_L: before sending to zero
04:09 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/arc_test2.png
04:09 PM Tom_L: and the 2 combined tests on the graphic
04:10 PM andypugh: OK. Bother!
04:10 PM seb_kuzminsky: no good deed goes unpunished
04:10 PM Tom_L: start and end points are met
04:10 PM andypugh: I doubt it is a real-world problem if any axis move prevents it.
04:11 PM andypugh: At a guess the initial XYZ are not _exactly_ zero.
04:11 PM seb_kuzminsky: the interference pattern of these conversations tells me we are missing tests of basic things like G2
04:12 PM andypugh: This would be easy enough to test, I think, as the missed-arc will take rather less time,
04:13 PM andypugh: Can someone add a debug print to _psoemath.c at 1849 of what the actual value of d is when it skips the arc?
04:13 PM Tom_L: i can also verify the helix works because i single point threads with it
04:13 PM andypugh: _posemath.c, I mean
04:13 PM Tom_L: but that's on 2.8
04:13 PM Tom_L: 2.8.4 iirc
04:14 PM seb_kuzminsky: i'm gearing up to do this test on 2.8 in a buster vm
04:14 PM andypugh: The large-arc fix which introduced this bug is in 2.8
04:14 PM andypugh: FWIW I can’t reproduce it in a Buster VM….
04:14 PM Tom_L: actually i may have not cut a helix since upgrading to 2.8.4
04:14 PM seb_kuzminsky: my repro was on bullseye
04:15 PM Tom_L: bookworm / master on the sim test above
04:15 PM andypugh: Tom_L: The bug was (probably) introduced in Feb this year.
04:16 PM Tom_L: i can try that on the mill if you want if it's within it's bounds
04:16 PM Tom_L: i'm set up for inch on it
04:17 PM seb_kuzminsky: my testing is with sim/axis/axis, which is inch
04:17 PM pere: seb_kuzminsky: <URL: https://github.com/petterreinholdtsen/linuxcnc/tree/test-classicladder-estop > is as far as I get with classicladder testing so far.
04:18 PM * Tom_L puts on a coat
04:19 PM seb_kuzminsky: the problem goes away if there's any machine motion before the first G2
04:19 PM seb_kuzminsky: if i `g0 x1` and `g0 x0` first, the problem doesnt happen
04:20 PM pere: uninitialized value somewhere, perhaps?
04:20 PM pere: perhaps a ignored compiler warning?
04:22 PM seb_kuzminsky: pere: that looks promising!
04:22 PM seb_kuzminsky: you should gitignore the .var file
04:22 PM pere: is there a way to stop linuxcnc from creating a .var file?
04:23 PM pere: seb_kuzminsky: I suspect I am on the right track, but do not really know if I am poking the right pins.
04:24 PM seb_kuzminsky: and i think you have a potential race in your python "ui", after you poke ext-estop-button you should call a function to wait a few seconds for linuxcnc to do the thing
04:24 PM pere: as in c.wait_complete()?
04:24 PM seb_kuzminsky: something like the "wait_for_*()" functions in linuxcnc_util.py
04:24 PM seb_kuzminsky: c.wait_complete() waits for the command to be sent to the motion controller
04:25 PM pere: not quite sure what to wait for...
04:25 PM seb_kuzminsky: i meant something different:
04:25 PM seb_kuzminsky: after you set yout hal pin that's connected to classicladder, you don't know how long it will take for classicladder to notice & react to it & the output to change
04:26 PM seb_kuzminsky: so you should poll hal.get_value('estop-all-ok') for a few seconds, not just fail the test if it's not immediately what it should be
04:27 PM pere: adding sleep 5 did not help
04:27 PM seb_kuzminsky: on a real realtime system it *should* be so quick that the next python line sees the effect, but in our dodgy testing infrastructure with crappy VMs and things it isn't always that good
04:28 PM seb_kuzminsky: ah, i haven't tried to run it yet, just giving feedback based on reading the patch
04:29 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/arc_test_2.8.4.png
04:29 PM Tom_L: took a bit
04:29 PM Tom_L: must be before the fix
04:30 PM seb_kuzminsky: Tom_L: that looks different - not the bug andypugh's chasing
04:31 PM pere: seb_kuzminsky: pushed patch, get past first test, fail on error: estop not OK aftet ext-estop-button is unset
04:31 PM seb_kuzminsky: on my buster 2.8 i don't see the problem
04:31 PM pere: look like I can get into estop, but not out of it.
04:32 PM seb_kuzminsky: there's a little song and dance you have to do to come out of estop
04:32 PM Tom_L: seb_kuzminsky, have you updated it to 2.8.4?
04:33 PM seb_kuzminsky: Tom_L: yeah, i'm on the tip of 2.8, 3e7c40e52ea0c
04:33 PM Tom_L: been a while since i updated the mill copy
04:34 PM Tom_L: a fix may have been pushed since
04:36 PM linuxcnc-build2: Build [#14](http://buildmaster-2.local:8010/#builders/6/builds/14) of `00-checkin` 4failed.
04:37 PM pere: seb_kuzminsky: perhaps estop is a bad test to do with classicladder, then?
04:37 PM seb_kuzminsky: pere: did you find the Ladder Editor? it's helpful for inspecting/debugging ladder stuff
04:38 PM Tom_shop: after sending the mill to zero, it tried to run the command
04:38 PM Tom_shop: but was beyond the limits
04:38 PM pere: seb_kuzminsky: I did, but did not tell me much. :/
04:39 PM seb_kuzminsky: after the machine enters estop, you need to ask linuxcnc to come out of estop again with STATE_ESTOP_RESET
04:40 PM seb_kuzminsky: both the machine (in this case, ladder + your python hal pin) and the controller have to agree before itwants to come out of estop, the
04:40 PM seb_kuzminsky: err
04:40 PM seb_kuzminsky: they have to agree before the machine will come out of estop
04:41 PM pere: seb_kuzminsky: right. seem to work. the test script now run, but I have no idea if I actually test classicladder.
04:42 PM pere: branch updated.
04:42 PM pere: horrible sleep(5) still in place.
04:43 PM Tom_L: andypugh's problem may not show on my mill since i home to a corner and must jog off it to run the test
04:43 PM seb_kuzminsky: Tom_L: good observation
04:46 PM seb_kuzminsky: we should really have better docs on how estop works
04:47 PM seb_kuzminsky: there are a bunch of pins involved, plus NML commands
04:48 PM seb_kuzminsky: motion.enable, io.emc-enable-in, io.user-enable-out, and io.user-request-enable
04:48 PM pere: my goal with this test is to toggle some pin that is input into classicladder, and look at some pin that is affected by this that originate from classicladder. I hope I manage to do so in this test, but am not sure.
04:50 PM seb_kuzminsky: maybe you want to make a different ladder setup, a simpler one that doesn't involve estop, for example the simple latching on-off circuit in the Ladder Introduction document
04:50 PM pere: I would love a simpler ladder setup... :)
04:50 PM pere: I picked estop because it was already in the example file I found.
04:51 PM pere: I could try the lube stuff, but I understand less of it than the estop part.
04:51 PM JT-Shop2: pere, ladder logic is a funny beast, inputs are read, logic solved, outputs updated and the last one wins
04:51 PM * JT-Shop2 is a PLC programmer among other things
04:51 PM seb_kuzminsky: i suggest making a ladder explicitly for testing, without having it connected to any of linuxcnc's core functionality
04:52 PM seb_kuzminsky: test it in isolation
04:52 PM pere: all I want is a test to verify classicladder keep working after <URL: https://github.com/LinuxCNC/linuxcnc/pull/1657 > is applied.
04:52 PM seb_kuzminsky: Tom_L: yeah, after jogging the problem goes away
04:53 PM JT-Shop2: pere, how about a self reseting timer?
04:53 PM Tom_L: seb_kuzminsky, so unless you home to the middle of your xy bounds you may never see the issue
04:53 PM JT-Shop2: https://gnipsel.com/linuxcnc-old/ladder/index.html
04:54 PM Tom_L: not many machines i know of do that
04:54 PM seb_kuzminsky: Tom_L: well, if you home somewhere where the problematic helix will fit, yeah
04:55 PM * JT-Shop2 retires to the couch
04:55 PM seb_kuzminsky: Tom_L: well, if you home somewhere where the problematic helix will fit, yeah
04:55 PM Tom_L: yes
04:55 PM Tom_L: realistically most commercial machines home to a corner
04:55 PM seb_kuzminsky: and a partial helix (g3 x2 y2 z-1 j2 f1000) doesn't trigger the problem here, though it would work from the corner
04:57 PM seb_kuzminsky: hm, actually g3 doesn't trigger it at all here, only g2, wtf
04:57 PM pere: JT-Shop2: any simple one with an input, an output and some ladder logic in between would work for me]
04:57 PM Tom_L: it's just for the lefthanded folks
04:58 PM Tom_L: i get a good G3 arc as well
05:00 PM Tom_L: seb_kuzminsky, overlay them on the same sim and you will see something interesting
05:01 PM -!- #linuxcnc-devel mode set to +v by ChanServ
05:01 PM Tom_L: seb_kuzminsky, http://tom-itx.no-ip.biz:81/~webpage/cnc/arc_test3.png
05:02 PM Tom_L: but that may be affected by G90 G91
05:04 PM pere: it is getting late, so I'll call it a day. see you tomorrow.
05:12 PM JT-Cave: damn fonts are really screwed up in debian 12 now
05:13 PM JT-Cave: I think it's a gtk issue dunno but almost unusable now
05:13 PM linuxcnc-build2: Worker `checkin` is missing. It was seen last on Sat Nov 26 15:11:39 2022.
05:14 PM Tom_L: seb_kuzminsky, andypugh it only happens with G90 active
05:14 PM Tom_L: G20
05:14 PM Tom_L: G91
05:15 PM Tom_L: G2 x0y0z-1j4 f1000 works as expected
05:17 PM Tom_L: double checked to confirm
05:17 PM JT-Cave: what's the cause of the issue?
05:18 PM seb_kuzminsky: pere: you might have seen this already, but i found this video helpful and informative to getting started with the Ladder Editor: https://www.youtube.com/watch?v=jkIk8c7axxQ
05:19 PM JT-Cave: and I thought my tutorial was helpful...
05:19 PM seb_kuzminsky: link?
05:19 PM seb_kuzminsky: i just clicked on the first one i found on youtube
05:20 PM JT-Cave: 04:53 PM JT-Shop2: https://gnipsel.com/linuxcnc-old/ladder/index.html
05:21 PM seb_kuzminsky: ah, yeah that's good!
05:21 PM JT-Cave: :)
05:21 PM seb_kuzminsky: i needed some basic hints on how to use the ladder editor gui - i didn't realize you couldn't place inputs & outputs first, and connect them later
05:22 PM Tom_L: so the error occurs in absolute mode (G90) only
05:22 PM seb_kuzminsky: you have to build the rungs up from the rails
05:23 PM JT-Cave: each section is on rung but it can be complex
05:24 PM seb_kuzminsky: yeah
05:24 PM JT-Cave: on one
05:28 PM Tom_L: <andypugh> _posemath.c, I mean
05:28 PM Tom_L: andypugh, if you tell me how...
05:30 PM JT-Cave: wow cloning mesact is slow tonight\
05:34 PM Tom_L: JT-Cave, must be on their end
05:36 PM JT-Cave: prob a busy time for starlink
05:36 PM Tom_L: mine was slow too
05:37 PM Tom_L: 40% then nothing
05:37 PM Tom_L: stopped..
05:37 PM Tom_L: 15.6 MiB/s
05:39 PM JT-Cave: 350 KiB/s
05:39 PM JT-Cave: gtk is fucked up on debian 12 atm
05:39 PM Tom_L: mine started at 443.7 MiB/s
05:40 PM Tom_L: is that the one that lost the gui?
05:40 PM JT-Cave: this pc lost the menu after I noticed some crap ide on programs menu and I tried to delete
05:40 PM Tom_L: err menu..
05:40 PM JT-Cave: it
05:43 PM JT-Cave: thought I saw a mouse in the run but it's a bug flying around
05:43 PM JT-Cave: I think
05:48 PM * JT-Cave calls it a day
05:53 PM andypugh: Tom_L: I had a play, and I haven’t figured out how to print from _posemath.
05:53 PM andypugh: There is a lot of IFDEF PM_PRINT_ERROR bit it seems to be specifically turned off with RTAI
05:53 PM Tom_L: you see the note about G90 G91?
05:54 PM andypugh: ?
05:54 PM Tom_L: i tested both modes
05:54 PM Tom_L: G90 fails the test, G91 passes
05:54 PM Tom_L: <Tom_L> G20
05:54 PM Tom_L: <Tom_L> G91
05:54 PM Tom_L: <Tom_L> G2 x0y0z-1j4 f1000 works as expected
05:54 PM Tom_L: <Tom_L> double checked to confirm
05:55 PM andypugh: Does that make sense>
05:55 PM andypugh: ?
05:55 PM Tom_L: not to me
05:56 PM Tom_L: the numbers would be larger in G90 absolute mode over incremental
05:56 PM andypugh: The test is definitely assuming G91
05:56 PM Tom_L: G91 worked here
05:56 PM Tom_L: G90 failed
05:56 PM andypugh: But if you are at 0,0,0 then the two modes are the same.
05:57 PM Tom_L: yes i suppose
05:57 PM Tom_L: i didn't send to zero prior to the tests
05:57 PM Tom_L: when i did it passed
06:00 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/arc_test3.png
06:00 PM Tom_L: also note what G2 did in G90 there
06:00 PM andypugh: In my testing, starting at 0,0,0 both G90 and G91 make a spiral. This is really confusing.
06:00 PM Tom_L: yes i agree with your tests
06:01 PM Tom_L: and i can't test on the mill since i home at a corner and have to move off it to run the test
06:01 PM Tom_L: only in sim and only in G90 mode does it fail
06:02 PM Tom_L: unless you home a machine where the arc will fit within the limits
06:02 PM Tom_L: ie middle of the table
06:08 PM Tom_L: ok i can make it fail in G91 also
06:12 PM Tom_L: G90
06:12 PM Tom_L: G2 X0Y0Z-1 F1000 <--fails
06:12 PM Tom_L: G91
06:12 PM Tom_L: G2 X0Y0Z-1 F1000 <--makes an arc
06:12 PM Tom_L: G2 X0Y0Z-1 F1000 <--fails
06:13 PM Tom_L: still in G91 mode
06:13 PM Tom_L: and by 'fails' i mean just a linear move in Z
06:15 PM Tom_L: andypugh, that sequence shown here: http://tom-itx.no-ip.biz:81/~webpage/cnc/arc_test4.png
06:22 PM andypugh: Tom_L: I think you are testing something else? G2 without a J?
06:33 PM linuxcnc-build2: Worker `checkin` is missing. It was seen last on Sat Nov 26 16:32:07 2022.
07:04 PM Tom_L: andypugh, no i just forgot to add the j in the irc chat.
07:04 PM Tom_L: check the left of the png
07:04 PM Tom_L: j4
07:06 PM andypugh: Well, I still get a circle…
07:06 PM Tom_L: i just posted the exact sequence (less the J) here that you see in the pic
07:07 PM andypugh: Are you homing, or is this a config that doesn’t force homing?
07:07 PM Tom_L: homing in the sim yes
07:08 PM Tom_L: the sim won't run until homed
07:08 PM andypugh: I get a spral every time.
07:08 PM Tom_L: the standard axis sim
07:09 PM andypugh: I am running this in a VM on my Mac (Intel).
07:09 PM andypugh: What is your CPU?
07:09 PM Tom_L: i5 intel
07:09 PM Tom_L: on bookworm
07:10 PM Tom_L: how do i check that in linux?
07:10 PM -!- #linuxcnc-devel mode set to +v by ChanServ
07:12 PM Tom_dev: Intel(R) Core(TM) i5-6600 CPU @ 3.30GHz
07:19 PM andypugh: My Mac has a Core i5 3GHz so it’s almost certainly not that.
07:41 PM Tom_L: andypugh, do you get a circle or a helix?
07:42 PM andypugh: I always get a circle or a helix (depending on the starting point) I have never seen a straight Z move
07:42 PM Tom_L: ok
07:43 PM Tom_L: i've tried several combinations and get mixed results depending on G90 G91
07:43 PM Tom_L: but still get the line
07:48 PM andypugh: I wonder if this is an uninitilaised memory problem and the VMS always start with zeroed memory?
07:49 PM andypugh: I will try a real PC tomorrow, but it’s roo late now
07:50 PM Tom_L: yep
07:50 PM Tom_L: holler if i can test something
07:50 PM Tom_L: putting it in a ngc file to run instead of mdi...
07:51 PM Tom_L: will post the file and png
07:53 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/arc_test/arc_testrun1.png
07:53 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/arc_test/arc_test.ngc
07:54 PM Tom_L: the preview shows it correct
08:00 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/arc_test/arc_testrun2.png
08:00 PM Tom_L: same file with G3 instead
08:07 PM andypugh: OK, sleep time
08:07 PM Tom_L: later
08:44 PM Tom_L: andypugh, an update.. i ran the same test file on an Rpi4 axis sim with the same results
08:51 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/arc_test/arc_test_Rpi4.png