Back
[10:08:29] <KGB-linuxcnc> 03Dewey Garrett 05joints_axes12 3cd7b92 06linuxcnc 10src/emc/usr_intf/axis/scripts/axis.py axis.py onoff_clicked force joint mode #45 'On sim' * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=3cd7b92
[10:09:18] <dgarr> skunkworks: i think this comit fixes the sim example of #45, can you test please
[10:11:30] <jepler> dgarr: my original impulse was to look for a place in task/motion to fix this. can you say something about why do it in each UI instead?
[10:11:32] <dgarr> (i git bisected to isolate, spent hours working on changes in motion/, then tested and found problem did not occur in other guis (tklinuxcnc))
[10:13:50] <dgarr> i think problem relates to the axis gui's insistence on hiding joints/axes distinctioins from users for identity kins, i understand the reasonihg but it makes everything very complicated in ja
[10:15:28] <jepler> I agree, in general changes in axis should be made to make machines with nontrivial kinematics work properly.
[10:19:04] <jepler> I am still concerned about how feedback cartesian position is not tracking motor feedback position in machine off state. Is it your *ahem* position that this is expected behavior?
[10:19:19] <jepler> .. and if you want to see that position changing in machine off state, you switch to joint mode?
[10:20:42] <dgarr> i have no position, if you or anyone else can fix/improve it is fine with me -- it is beyond by capabilities
[10:23:32] <jepler> thanks for taking the time to discuss it with me.
[10:23:43] <jepler> it seems like this will certainly help skunksleep in the short term
[10:28:57] <jepler> well I put my thoughts on the bug as well
[10:29:13] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15jepler commented on issue #45: I think another connected issue is how ja doesn't update cartesian feedback position from joint feedback position while in "machine off" mode. ... 02
https://github.com/LinuxCNC/linuxcnc/issues/45#issuecomment-195428921
[10:31:41] <dgarr> jepler: i think thechange i made to onoff_clicked should be done only for IDENTITY, do you agree?
[10:42:56] <skunkworks> dgarr, thanks - that seems to have fixed it. I will try looking at your config for the jogging issues I have having. (following errors)
[11:05:15] <jepler> dgarr: I don't know.
[11:12:59] <dgarr> it does cause unexpected behavior for non-identity kins (ldelta.ini for example) so i will change onoff_clicked
[11:13:44] <KGB-linuxcnc> 03Dewey Garrett 05joints_axes12 3452fef 06linuxcnc 10src/emc/usr_intf/axis/scripts/axis.py axis.py onoff_clicked joint mode iff identity * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=3452fef
[12:39:26] <jsskangas> https://www.irccloud.com/pastebin/yy2neBXk
[12:57:43] <cradek> halui is not realtime
[13:11:13] <jsskangas_> hello
[13:11:47] <jsskangas_> So I got this IO-card runnig with comedi driver as userspace component
[13:12:53] <jsskangas_> but Im having a add problem, signals from io-card is connected to HALUI to axis jog signals plus and minus
[13:13:31] <jsskangas_> probplem is I gettin sticky buttons and lag from bushing button
[13:13:50] <jsskangas_> compared to actual axis jog movements
[13:14:16] <jsskangas_> but looking these signal from halscope there is no lag or stickyness
[13:15:33] <jsskangas_> I made video when jogging X axis it show halui jog pin, comedidriver pin that is connected to pin jog speed command and axis pos
[13:16:06] <jsskangas_> Jog speed somehow is reletad to this lag an stickyness
[13:16:32] <jsskangas_> I will post link to this video soon
[13:16:39] <jsskangas_> as its get uploaded
[13:17:37] <jsskangas_> Any ideas what could cause this
[13:17:46] <jsskangas_> ?
[13:20:00] <jepler> for jogging, it is preferable to use the realtime inputs to motion rather than the non-realtime inputs to halui
[13:21:04] <jepler> this has more non-realtime bits which can all introduce their own delays: [non-realtime component] - hal pins - [halui] - nml - [task] - shared memory - [realtime motion]
[13:21:13] <jepler> vs [non-realtime component] - hal pins - [realtime motion]
[13:22:19] <jepler> man 9 motion for a reference of pins that motion exports, including motion.* and axis.*. This includes the ability to do activities like jog, but not the ability to do things like change the machine mode like estop/on/off or manual/mdi/auto, so for those you will still need to involve halui
[13:29:22] <jsskangas_> ok jogging pin to motin not to halui?
[13:29:34] <jsskangas_> I will try that this night
[13:44:30] <jsskangas_> jepler so basicly this means that i need to some how chanke +/- button signal to jog counts?
[13:45:46] <jepler> yes, I guess so.
[13:46:02] <jepler> I don't know whether the realtime inputs to motion have a continuous jog option
[13:46:06] <jsskangas_> how its done from axis
[13:46:20] <jsskangas_> from keyboard?
[13:46:23] <jepler> [axis] - nml - [task] - shared memory - [realtime motion]
[13:46:37] <jsskangas_> I just need soimilar functionality
[13:46:42] <jepler> of course the X window server, keyboard/mouse driver, etc, are all non-realtime and before axis
[13:46:59] <jepler> what is your base version of linuxcnc?
[13:48:59] <jsskangas_> 2.6.something
[13:49:01] <jsskangas_> https://www.youtube.com/watch?v=4d0Cn98bVCA
[13:49:16] <jsskangas_> here is video about this
[13:49:47] <jsskangas_> note that buttons give signal every time
[13:50:10] <jepler> upgrade to 2.7. we made important improvements in how multiple UIs communicate with task using nml
[13:50:21] <jepler> any 2.7.x version will have these improvements.
[13:50:38] <jsskangas_> I just need good way to connect these signals to jos axis
[13:51:05] <jsskangas_> and easyestr I found was halui jog minus and plus pins
[13:51:13] <jepler> "nml: implement command queue with reliable reception"
[13:51:19] <jsskangas_> just what I need but this lag....
[13:51:43] <jepler> I strongly suspect the lag will be better once you upgrade to a current version
[13:52:06] <jepler> now that I know you are using 2.6 I am not surprised that you are having problems of this nature.
[13:52:09] <jsskangas_> but why keyaboard has no such lag
[13:52:13] <jepler> I belive it's much better in the current version
[13:52:39] <jsskangas_> its not about userspace component
[13:52:49] <jsskangas_> you can see that from video
[13:52:50] <jepler> because halui and axis are two separate programs both communicating with task via nml
[13:53:02] <jsskangas_> ok
[13:53:30] <jepler> we made improvements of that scenario in particular.
[13:53:44] <jsskangas_> ok really good to know
[13:53:46] <jepler> in 2.6 and earlier, each UI "polls" task to find out if its command has been received. this is unreliable and creates delays
[13:53:58] <jsskangas_> will 2.7 run on 10.04 ubuntu
[13:54:00] <jsskangas_> ?
[13:54:11] <jsskangas_> is there any restrictions
[13:54:24] <jepler> according to
http://wiki.linuxcnc.org/cgi-bin/wiki.pl?MinimumSoftwareVersions yes
[13:56:51] <jsskangas_> OK, I try to upgrade to 2.7
[13:58:07] <jsskangas_> any documentation for updating
[13:58:08] <jsskangas_> ?
[13:58:23] <jepler> http://linuxcnc.org/docs/2.7/html/getting-started/updating-linuxcnc.html
[14:00:11] <jsskangas_> thank
[14:00:47] <jsskangas_> its bottle of good rum and then try that
[14:00:49] <jsskangas_> thanks
[14:08:17] <jepler> more background about the multiple user interfaces over nml problem in 2.6 and earlier, and which we improved in 2.7:
http://mid.gmane.org/20141020115421.GB32102%40unpythonic.net https://sourceforge.net/p/emc/bugs/328/
[14:12:21] <jsskangas> I will check that
[14:20:50] <skunkworks_> zlog
[14:48:41] <automata_> hi... we are looking to tackle display of line numbers and file names while the auto file is running problem
[14:49:12] <automata_> found this document:
http://wiki.linuxcnc.org/cgi-bin/wiki.pl?LineNumbers
[14:49:29] <automata_> is the document still relevent?
[14:55:27] <jepler> I don't know.
[15:06:29] <automata_> from what I am seeing, source context for the motion ID being executed is not available at run time.
[15:07:17] <automata_> so I am trying to put in some effort to get the exact source file and line number for the currently executing file
[15:07:31] <automata_> currently executing line
[15:08:27] <cradek> automata_: you might want to check out the state-tags-master branch, where a lot more information is passed along with the motions
[15:08:43] <automata_> any guidance on the issue will be much appreciated
[15:08:47] <cradek> it hasn't been merged because it has some problems:
https://sourceforge.net/p/emc/bugs/search/?q=labels:"state-tags-master"
[15:09:16] <cradek> also I don't know whether it passes a more-detailed file/line context
[15:09:20] <automata_> looking at it right now
[15:10:17] <automata_> today we deciphered how line numbers are passed to the TP / TC via the interp_list and NML message
[15:15:32] <cradek> that in itself is an accomplishment
[15:17:56] <automata_> i added a source context which is accessible in the executor and can log the line numbers and file numbers as an auto file is executing
[15:22:11] <automata_> seems like rob has done a lot of work on state tags....
[15:22:42] <automata_> any idea why it was not cmpleted / abandoned?
[15:28:10] <cradek> just because of those bugs, as far as I know. tormach has said a few times they'd look into them, but they must be busy elsewhere.
[15:28:29] <cradek> if you'd like to work on it, that would be wonderful
[16:30:58] <micges> logger[psha]: hi
[16:30:58] <logger[psha]> micges: Log stored at
http://psha.org.ru/irc/%23linuxcnc-devel/2016-03-11.html
[23:24:12] <KGB-linuxcnc> 03Chris Morley 052.7 d43a5f5 06linuxcnc 10src/emc/usr_intf/pncconf/pncconf.py 10src/emc/usr_intf/pncconf/screen.glade pncconf: fix gui's jog default settings * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=d43a5f5
[23:27:48] <KGB-linuxcnc> 03Chris Morley 05master 2957cc5 06linuxcnc 10lib/python/pyngcgui.py Merge branch '2.7' * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=2957cc5