#linuxcnc-devel | Logs for 2016-03-09

[02:10:15] <cmorley> The gmoccapy py option is stepconf is suboptimal ... there are defaults that should be handled for a better user experience. such as GLADE panels...Gmoccapy can't use pyvcp panels... so there is still work that can be done. Norbert could advise
[02:11:10] <cmorley> yland_ ^^^
[02:11:17] <cmorley> Night
[08:35:51] <jepler> seb_kuzminsky: yay! ja12
[08:40:36] <jepler> and of course dgarr yay ja12
[10:43:43] <seb_kuzminsky> alphago 1, sedol 0
[10:43:47] <seb_kuzminsky> it's the end of an era
[11:20:51] <jepler> I'll wait until at least the second game to give up on humanity
[11:21:18] <jepler> but I think we should take heart that skynet is content to beat us at go, and not commit genocide as predicted in movies.
[11:22:12] <pcw_home> so far...
[11:51:56] <skunkworks__> llog
[11:51:59] <skunkworks__> zlog
[11:54:31] <skunkworks__> I am trying to get ja12 running on the K&T and a couple issues. joint.4.unlock and joint.4.is_unlocked don't exist. I remarked them out so I could get linuxcnc to load but now am getting
[11:54:50] <skunkworks__> Traceback (most recent call last): File "/bin/arduino", line 56, in <module> c['analog-in-%02d-gain' % port] = 1.0 SystemError: null argument to internal routine ./hm2-servo.hal:879: waitpid failed arduino arduino ./hm2-servo.hal:879: arduino exited without becoming ready
[11:55:19] <seb_kuzminsky> the arduino problem is not ja's fault
[11:55:38] <seb_kuzminsky> the joint pins missing might be, or maybe a bug in the update_ini script?
[11:55:41] <skunkworks__> the arduino stuff works on the installed version of linuxcnc (22.7.4)
[11:55:45] <skunkworks__> 2.7.4
[11:56:21] <seb_kuzminsky> hmm, can you paste the converted hal file?
[11:57:54] <skunkworks__> I could remark it all out...
[11:57:57] <skunkworks__> hold on
[11:59:01] <jepler> skunkworks__: it would be good to discover whether the arduino script you are using is broken in master, or just in ja.
[11:59:07] <jepler> it seems unlikely for it to be ja's fault
[12:01:39] <skunkworks__> http://electronicsam.com/images/KandT/conversion/testing/config/KandTzeroJA12.tar.gz
[12:02:29] <jepler> I bet the most likely commit to have caused breakage is 2b6c8ad7102
[12:03:45] <jepler> >>> import hal
[12:03:46] <jepler> >>> c = hal.component("bob")
[12:03:48] <jepler> >>> c.newparam("g", hal.HAL_FLOAT, hal.HAL_RW)
[12:03:50] <jepler> >>> c['g'] = 1.0
[12:03:52] <jepler> SystemError: null argument to internal routine
[12:03:59] <jepler> I'll file a github issue
[12:04:08] <seb_kuzminsky> great
[12:04:15] <skunkworks__> that is the converted ini/hals.. Also the conversion script used to run automatically.. It doesn't now? http://imgur.com/1pHmDcz
[12:04:55] <skunkworks__> Thank you!
[12:06:58] <jepler> skunkworks__: do you have a github account?
[12:08:14] <skunkworks__> I do not yet. Linuxcnc launched - now to see why the locking pins are not there
[12:08:30] <skunkworks__> The B axis is.
[12:08:32] <seb_kuzminsky> skunkworks__: the pin is named ".is-unlocked", not ".is_unlocked"
[12:08:34] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15jepler opened issue #43: python hal module broken badly in master branch 02https://github.com/LinuxCNC/linuxcnc/issues/43
[12:08:59] <jepler> skunkworks__: ok. if you did, you could subscribe to that issue but you'll hear about it here if I get around to closing it so nbd
[12:09:17] <jepler> .. but I can't work on it right now. ping me, the breakage is almost certainly my fault.
[12:09:36] <seb_kuzminsky> looks like motion needs an "unlock_joints_mask" argument now, but doesn't get it from the ini file
[12:10:19] <seb_kuzminsky> in master, axes 3-5 (the rotaries) all get the locking pins always
[12:10:29] <seb_kuzminsky> i'll open a github issue for that one
[12:12:07] <skunkworks__> the lock pins don't exist in ja it looks like
[12:13:23] <skunkworks__> or does LOCKING_INDEXER = 1 need to be in the joint section?
[12:13:44] <seb_kuzminsky> skunkworks__: try changing your hm2-servo.hal loadrt line that loads EMCMOT, add "unlock_joints_mask=0x10"
[12:16:22] <skunkworks__> That did it
[12:17:39] <skunkworks__> I can't start the K&T at the moment.. Baby sleeping
[12:17:50] <skunkworks__> see then if it homes and such
[12:18:52] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15SebKuzminsky opened issue #44: ja: motion's joint locking hal pins missing 02https://github.com/LinuxCNC/linuxcnc/issues/44
[12:19:18] <seb_kuzminsky> the K&T has sure found a lot of bugs :-)
[12:21:08] <skunkworks__> :)
[12:21:36] <skunkworks__> It was a long time running master.. now it runs released perfectly
[12:23:04] <seb_kuzminsky> time to break it again with a bunch of experimental code :-)
[12:32:06] <KGB-linuxcnc> 03Jeff Epler 05master 78e6972 06linuxcnc 10src/hal/halmodule.cc halmodule: Fix 'SystemError: null argument to internal routine' * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=78e6972
[12:32:14] <seb_kuzminsky> heh
[12:32:25] <jepler> yay for getting something done in the first 5 minutes of lunch
[12:33:05] <seb_kuzminsky> i swear ^V is responsible for more bugs than any other ascii code
[12:37:52] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15cradek closed issue #43: python hal module broken badly in master branch 02https://github.com/LinuxCNC/linuxcnc/issues/43
[12:39:38] <skunkworks__> Do i need to test master?
[13:11:01] <seb_kuzminsky> skunkworks__: that wouldn't hurt
[13:11:40] <skunkworks__> in the process..
[13:12:47] <skunkworks__> did you see my question about the conversion utility?
[13:14:59] <skunkworks__> if you run an old config from the config picker - you get the above error.
[13:15:07] <skunkworks__> in ja
[13:21:03] <skunkworks__> jepler: seb_kuzminsky: master launches with the arduino loading - thanks
[13:41:07] <seb_kuzminsky> skunkworks__: did you get that update_ini usage popup when you tried to run ja12 on an un-converted 2.7 config?
[13:50:10] <skunkworks__> yes
[13:50:41] <skunkworks__> I converted it using the update_ini from command line then
[14:08:57] <seb_kuzminsky> bummer
[14:17:01] <seb_kuzminsky> skunkworks__: how did you launch linuxcnc? did you name your ini file ("linuxcnc ../configs/5i20.ini"), "linuxcnc -l", or just "linuxcnc"?
[14:27:20] <skunkworks__> just linuxcnc
[14:27:56] <skunkworks__> if I recall - it would just popup saying it was going to convert the files
[14:29:29] <skunkworks__> Hmm - the b axis doesn't home...
[14:32:10] <skunkworks__> Ok - joint.4.homed turns true - just doesn't show in axis
[14:33:46] <skunkworks__> The second I try to jog I get a following error
[14:34:58] <skunkworks__> It also doesn't ask if you are 'sure' you want to home - it just homes again
[14:35:16] <skunkworks__> maybe because axis doesn't know it is homes
[14:35:18] <skunkworks__> homed
[14:36:41] <skunkworks__> yes - axis says that it cannot move without being homed (doing a mdi command)
[14:40:54] <skunkworks__> So - axis doesn't know it is homed and for some reason jogging an axis with the jog wheel causes a following error
[14:44:03] <skunkworks__> The ini constraints look right..
[14:49:11] <skunkworks__> continuous jogging works
[14:56:12] <dgarr> skunkworks__: the update_ini doesn't handle locking indexers, instructions for the differences are given at http://buildbot.linuxcnc.org/doc/scratch/v2.8.0~pre1-ja~joints-axes12~bd31d87/html/getting-started/updating-linuxcnc.html#_locking_rotary_indexer_updates_for_joints_axes
[14:56:43] <dgarr> their is also a sim config for comparison: configs/sim/axis/axis_9axis.ini which uses B as a indexer
[14:56:47] <dgarr> bbl
[15:01:57] <skunkworks__> I added LOCKING_INDEXER_JOINT = 4 to the ini under the B axis (everything else was there) Could it be the issue with there being a joint 3 that is unused?
[15:02:12] <skunkworks__> it still doesn't home axis B in axis
[15:27:50] <KGB-linuxcnc> 03Jeff Epler 05master 6e895a0 06linuxcnc 10docs/src/getting-started/updating-linuxcnc.txt docs: be pedantic * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=6e895a0
[15:45:43] <skunkworks__> I can home the b axis by doing a Machine->homing-> home axis B. This then shows the homing target in axis - but wheel jogging still causes instant following errors and but i can command movement through mdi which seems to work
[15:54:35] <cradek> has a wheel jog ever unlocked?
[15:54:51] <cradek> er, can you jog a locked axis at all? I don't remember how it works
[15:57:58] <seb_kuzminsky> skunkworks__: so wait, you have axes XYZB, but you have joints 0, 1, 2, and 4? i thought joints were supposed to be 'compact', with no missing ones like 3 in your case
[15:59:21] <skunkworks__> Well - I just took the config and converted it...
[15:59:56] <skunkworks__> Isn't it how the current system works? B is 4
[16:00:39] <skunkworks__> cradek: not jogging the b axis - just thinking that the B axis not being homed was causing some odd issue
[16:01:11] <skunkworks__> I can try to change the config so it is just xyzb 0123
[16:05:35] <seb_kuzminsky> yeah i think that's an issue with the update script - or with my understanding of how joint numbers work in the ja branch
[16:07:52] <skunkworks__> Oh - that made it so b homed as expected.
[16:08:00] <skunkworks__> Yay
[16:08:24] <skunkworks__> But wheel jogging is still broken
[16:09:22] <skunkworks__> how do you switch to joint mode?
[16:09:58] <cradek> $
[16:10:01] <skunkworks__> exceeded positive soft limit - hint switch to joint mode and jog off soft limit
[16:10:38] <skunkworks__> I don't think that works with trivkins
[16:11:09] <cradek> oh hmm
[16:11:34] <cradek> maybe the message is misleading
[16:11:42] <cradek> how did you exceed the soft limit in the first place?
[16:12:00] <skunkworks__> I think the wheel jogging did it..
[16:12:54] <skunkworks__> * the wheel jogging did it - I clicked the wheel towards the home position and it somehow overshot
[16:13:19] <skunkworks__> now - where is the crowbar to release the y brake ;)
[16:14:01] <cradek> hmm
[16:14:12] <cradek> wheel jog sure shouldn't be able to jog past soft limits
[16:14:34] <cradek> with trivkins I wonder if you get joint wheel jog or axis wheel jog
[16:15:19] <skunkworks__> wow - there are issues.
[16:15:40] <skunkworks__> I just slid the Y down and now it won't power back up because of a following error
[16:16:23] <cradek> maybe command isn't tracking feedback in machine-off like it's supposed to
[16:16:46] <skunkworks__> right
[16:17:24] <skunkworks__> yes - it still thinks it is past the soft limit
[16:18:54] <skunkworks__> http://imgur.com/gJtq0UO
[16:19:15] <skunkworks__> You can see the Y axis is lower than 36...
[16:20:19] <cradek> do you have a smaller safer servo machine to experiment on?
[16:20:28] <skunkworks__> heh - no.
[16:20:34] <cradek> darn
[16:21:05] <skunkworks__> nothing as complete
[16:21:13] <skunkworks__> (home switches and such_
[16:21:17] <cradek> yeah
[16:25:49] <skunkworks__> It is hooked to joints..
[16:29:52] <skunkworks__> Hmm changing it to axis.x from joint.0 and so on - I get no wheel jogging
[16:30:11] <cradek> that might be correct for trivkins
[16:30:34] <cradek> do you have joint limits (for wheel joint jogging to see) in the ini?
[16:31:09] <cradek> oh yeah I guess so, since it says that's the one you exceeded
[16:32:21] <skunkworks__> hold on.. I don't think I have all the places changed
[16:33:13] <skunkworks__> Hmm - still nothing
[16:38:00] <skunkworks__> I think I may have to give up for the day
[16:38:53] <cradek> thank you for trying it
[16:40:18] <skunkworks__> No problem.
[16:40:24] <skunkworks__> I am good at breaking things
[16:41:50] <skunkworks__> current try http://electronicsam.com/images/KandT/conversion/testing/config/KandTzeroJA12new.tar.gz
[16:42:16] <jepler> in other big news, a new c/10 orthogonal spaceship in standard "life" rules has been discovered. https://niginsblog.wordpress.com/2016/03/07/new-spaceship-speed-in-conways-game-of-life/ maybe not as big as the go news
[16:42:58] <cradek> with config changes it's hard to tell what's a bug and what's misconfiguration (or also, possibly a bug that misconfiguration isn't caught and giving an error)
[16:57:15] <skunkworks__> sue
[16:57:17] <skunkworks__> sure
[17:01:05] <jepler> holy cow, scaleway has announced bare metal (not virtualized) x86-64, 4 cores, 8GB RAM, 50GB SSD at 12EUR/month
[17:01:18] <jepler> MHz/GHz of the CPUs not stated
[17:01:41] <jepler> and virtualized 2 cores, 2GB RAM, 50GB SSD for 3EUR/month
[17:03:49] <cradek> on a pipe, with a static IP?
[17:04:38] <jepler> yes
[17:04:55] <cradek> wow
[17:04:57] <jepler> commenters say the CPU is Atom C2750, which is not fast
[17:06:02] <jepler> 1 public IPv4, "unlimited transfer" 300MBit/s internet bandwidth
[17:06:19] <jepler> or 200Mbit/s on the 3EUR model
[17:11:28] <jepler> of course, never read the terms. "Users [...] are obligated not to...propagat[e] data, images or sounds that may constitute [...] denigration of [...] good morals or public order"
[17:23:57] <cradek> so "good morals" is defined elsewhere in the document?
[19:18:10] <jepler> of course not
[22:50:56] <cmorley> bug in the new trajectory planner? https://forum.linuxcnc.org/forum/38-general-linuxcnc-questions/30541-jerk-after-plunge-move-with-spindle-on?limitstart=0