#linuxcnc-devel | Logs for 2016-06-24

[04:15:46] <ikcalB> :bn
[04:16:22] <ikcalB> hey guys. tnx for all your support, especially jepler - the preview does run again.
[04:17:28] <ikcalB> it was taken (saying of my colleague) from another c/c++ gui 3y ago. unfortunately, it's not that usefuel, as it does take ages for any larger program to load
[04:18:12] <ikcalB> is there an easy possibility, to use gremlin (or whatever axis uses for preview - still investigating) in c++?
[07:09:33] <jepler> seb_kuzminsky: maybe a goto then
[07:11:47] <jepler> ikcalB: at best it would be a challenge to do. I'm not sure how you would achieve it.
[07:12:17] <jepler> seb_kuzminsky: oh wait a goto is your proposal.
[07:13:16] <ikcalB> jepler: had a look into the files, on would probably have to write a wrapper to use the gremlin implementation of python. yet I managed to make our c++only preview to suck much less.
[07:15:03] <ikcalB> your patch regarding `InterpBase *makeInterp()` works fine! the same would be needed for `knot_vector_creator` (used in NURBS_FEED)
[07:17:29] <jepler> isn't that prototype already in the header file canon.cc ?
[07:17:51] <jepler> er canon.hh
[07:18:19] <ikcalB> prototype is there, I got undefiend reference. implementation is not in any library (same for makeInterp, as ofc, this was not intended before)
[07:19:01] <jepler> oh sorry, it's too early in the morning and I have things backwards
[07:20:02] <ikcalB> to have things backwards? how early is it, @yours? we've got nice 30°C @ 2pm
[07:20:49] <jepler> oh I'm in the middle of the US, so it's 7AM and apparently just 18°C
[07:21:07] <jepler> 33°C later though
[07:21:50] <ikcalB> western AUSTRIA (no L ;) here. so grab a coffe, and relax. that was just fyi
[07:25:20] <jepler> I could find it on the map in only two tries
[07:26:10] <ikcalB> :D take that as a compliment!
[07:27:34] <jepler> well I'm still in the dark about your problem regarding knot_vector_creator though
[07:28:14] <jepler> let me try one more thing..
[07:28:45] <ikcalB> when linking against the package (libraries), there is the prototype in canon.hh. when viewing exported symbols with `nm`, I cannot find that functions beeing in any
[07:29:08] <ikcalB> I suppose that's why the linker cannot find a suiting implementation
[07:41:40] <jepler> jepler@rat:~/src/linuxcnc/master$ nm -C lib/librs274.so | grep knot_vector_creator
[07:41:43] <jepler> 00000000000fda0d T knot_vector_creator(unsigned int, unsigned int)
[08:03:39] <ikcalB> nm: /usr/lib/librs274.so: no symbols
[08:16:39] <ikcalB> btw: this should be a clean 2.7.4 from the reps
[08:30:32] <ikcalB> tried building again (as I was previously not linking against librs274.so) and it works. STILL idk what I'm doing wrong with the `nm -C /usr/lib/librs274.so` - any help?
[08:31:07] <ikcalB> leaving now, will have a look @zlog on monday at latest. have a nice weekend everybody
[09:24:48] <lair82_> Would anyone know when the next round of bug fixes will be available? like a 2.7.5 release?
[09:26:22] <lair82_> Norbert pushed a fix for a different issue than the one I brought up yesterday about the permissions problems, this relates to a problem with the icon to select and load the program selected from the nc_files folder not being activated.
[09:27:19] <lair82_> talked about here, https://forum.linuxcnc.org/forum/gmoccapy/30760-some-controls-not-functional-after-update
[09:27:58] <lair82_> The workaround for now, doesn't work on my ubuntu machine, it did on my debian though.
[10:03:57] <KGB-linuxcnc> 03Jeff Epler 052.6 a38665b 06linuxcnc 10src/emc/task/emctaskmain.cc task: Fix serial number handling after 516deaef * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=a38665b
[10:03:57] <KGB-linuxcnc> 052.6-task-fix a38665b 06linuxcnc 04. branch deleted * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=a38665b
[10:05:30] <seb_kuzminsky> lair82_: i had intended to make a stable release a few weeks back, but it was postponed due to bugs
[10:05:49] <seb_kuzminsky> i'm hoping the dust will settle from this task abort fiasco soon, then i'll make 2.6.13 and 2.7.5
[10:08:02] <lair82> Ok Seb, they just have to use the "Enter" key on the keyboard to select the highlighted file to make it work right now, just aggravating finding these issues, then have the workaround work on one system, then walk over to he next, and it not work.
[10:09:12] <lair82> Kinda like yesterdays problem with file permissions and gmoccapy's file editor problem.
[10:11:16] <jepler> norbert doesn't hang around on irc so this channel is not a good way to bring problems to his attention.
[10:11:47] <lair82> I have noticed this
[10:12:01] <lair82> I posted to the forum regarding this
[10:12:33] <lair82> I just don't know if I should use the forum, or hit directly through email
[10:13:30] <cradek> I think filing bug reports is much better than a forum thread, because we track them and know about them for coordinating bugfix releases
[10:14:19] <jepler> I would really discourage someone from reporting a bug in direct e-mail to the person they think should fix it, unless you already have a relationship with that person
[10:14:31] <lair82> Would that be a valid bug, the file select icon not working after an upgrade?
[10:14:38] <jepler> I basically ignore 99% of e-mails direct to me about linuxcnc, because they should be directed toward a public, community area instead
[10:16:59] <seb_kuzminsky> lair82_: if something isn't working as you expect it should, especially if it used to work, then that should be a bug report on github
[10:17:25] <seb_kuzminsky> it brings the issue to the developers' attention, and you can have a discussion with the relevant parties in the comments of the issue
[10:17:56] <seb_kuzminsky> if it turns out to be user error or miscommunication or something (ie not a bug in the code) we can just close the issue, no harm done
[10:18:14] <lair82> Ok, in past I steered away from that, because most times it was because I fouled something up, but I am pretty confident now with all the stuff I am doing with these now.
[10:18:26] <seb_kuzminsky> great!
[10:19:39] <jepler> If you prefer to take it to the forum first to get a second opinion (or to irc as you have today) that's great too
[10:22:36] <jepler> (and the presence of a workaround, particularly if it's nutty or doesn't work all the time, doesn't make something not a bug)
[10:24:41] <lair82_> The crazy thing is, i just used meld to compare the originals, to new file in master he said to use, and there is no difference between the files???
[10:27:37] <seb_kuzminsky> which file? which versions?
[10:28:17] <jepler> > If you do replace iconview.py with the file from master, all should be OK
[10:30:11] <jepler> there are 2 commits to iconview.py in the 2.7 branch since 2.7.4
[10:30:16] <jepler> *shrug* and afk
[10:34:42] <seb_kuzminsky> http://highlab.com/~seb/linuxcnc/latency-histogram-3.18-rtai.png
[10:34:50] <seb_kuzminsky> http://highlab.com/~seb/linuxcnc/latency-histogram-4.1-preempt-rt.png
[10:35:15] <seb_kuzminsky> tl;dr: < 70 us jitter with 4.1-preempt-rt
[10:35:33] <seb_kuzminsky> (i didn't test 4.6 because of the xorg dependency)
[10:35:49] <seb_kuzminsky> i vote we put 4.1-preempt-rt in our deb archive
[10:35:51] <seb_kuzminsky> bbl
[10:36:12] <cradek> sweeet
[10:41:40] <lair82_> Can't I just do a git pull, to update that file?
[10:41:58] <lair82_> And it should work?
[11:04:05] <lair82_> Got it, had to copy it to /usr/share/pyshared/gladevcp/ to get to work properly. But thats not where I put it on the debian machine this morning.
[11:04:47] <lair82_> I put the new copy in the /linuxcnc-dev/lib/python/gladevcp/ directory,
[11:04:57] <lair82_> on the debian machine
[11:14:04] <jepler> never manually copy files over installed files.
[11:15:01] <jepler> seb_kuzminsky: 4.1 works on both my systems (I've only posted results for one of them, but I know I ran 4.1 on my laptop in the past)
[11:15:07] <jepler> seb_kuzminsky: so yes +1 to that
[11:15:25] <jepler> also people who have followed pcw's instructions have 4.1, and I think that includes jt and skunkworks
[11:18:53] <jepler> if I wanted to run 4.1 myself with nvidia I'd have to sort through that headache, but that's my problem
[11:21:24] <lair82_> I try to leave well enough alone, but after I updated, and saw this, the shop supervisor would have had a shit fit, because it would have required an extra step hitting the enter key.
[11:27:22] <jepler> when installing a different version of the package, the package system is not going to make some attempt to integrate your changes, it's just going to throw them out
[11:32:15] <jepler> I haven't kept track of how debian packages Python but at one point there was something even worse you could manage to do to yourself:
[11:32:23] <jepler> the package would put a .pyc (byte-compiled python) file in place
[11:32:34] <jepler> if you took bad advice, you would copy a .py file into the same directory
[11:32:42] <jepler> Python will take the .py file and not the .pyc file, if they don't match
[11:32:52] <jepler> now, updating never gets you the correct (updated) file anymore
[11:33:02] <jepler> so just don't do it
[11:33:11] <jepler> if you need to make local mods then either use RIP or build and install your own .deb
[11:33:59] <jepler> in either of those cases, track your changes with git, and merge or rebase when new versions come out so that you can integrate your local changes with our changes; git will track our changes and your changes both
[12:43:34] <mozmck> It seems that the tool in gremlin does not scale with the size of the part - is there a way to make it do that?
[12:43:56] <jepler> yes, by editing the source code
[12:44:09] <mozmck> :-)
[12:44:52] <mozmck> ok - I didn't know if there was a setting or something I was missing.
[12:46:36] <jepler> I don't think so
[12:48:43] <mozmck> ok, thanks. I'll look at it. I had a question about it and opened a file that filled a 48" x 48" sheet and the tool is pretty large in comparison.
[12:50:57] <mozmck> jepler: would having the Wichita meeting a week sooner or later help for your schedule?
[12:52:33] <jepler> ending before August 3 or beginning after August 14 are both more likely to work for me
[12:55:04] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15geris2 opened pull request #83: Fixed some booting issue with GM6-PCI driver (062.7...06general-mechatronics-PCI-driver-fix) 02https://github.com/LinuxCNC/linuxcnc/pull/83
[12:55:35] <mozmck> Oh, well the proposed date was Oct 7 and forward.
[12:56:04] <jepler> yeah I mean October
[12:56:08] <jepler> I am just bad at saying the names of things
[12:56:19] <mozmck> I see - do it plenty myself.
[12:57:34] <mozmck> The week before would be fine with me, and Stuart indicated the same for him.
[12:59:04] <mozmck> You might post to the user list and mention that.
[12:59:23] <jepler> this would be more like 2 weeks before
[12:59:30] <jepler> wouldn't it?
[13:00:09] <mozmck> Looks like the 3rd is the monday of the proposed week
[13:00:54] <mozmck> after the 14th would be 2 weeks after.
[13:01:13] <jepler> right, I need to be at my day-job solid from the 3rd to the 14th
[13:02:00] <cradek> fwiw, I'm probably in the same boat
[13:03:07] <mozmck> Yeah, so if the meeting was the week before the weekend of 7,8,9, then it would be Sept 29,30, Oct 1, and maybe 2
[13:09:05] <jepler> that is a better possibility for me personally.
[13:12:02] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15SebKuzminsky commented on issue #83: Looks good to me. I'll merge it this afternoon. 02https://github.com/LinuxCNC/linuxcnc/pull/83#issuecomment-228412793
[13:18:00] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15jepler commented on issue #83: Proposed release note:... 02https://github.com/LinuxCNC/linuxcnc/pull/83#issuecomment-228414176
[14:00:04] <mozmck> Hmm, it would seem there is some sort of scaling issue in gremlin: http://pasteboard.co/1VB2KIKw.png
[14:00:11] <mozmck> The units are inches.
[14:00:40] <cradek> which units are inches?
[14:00:51] <mozmck> in the picture
[14:01:18] <mozmck> the "part" is about 48" x 96"
[14:01:43] <cradek> I don't understand what you are saying is wrong
[14:01:43] <mozmck> The text for the measurements is huge as is the tool display.
[14:01:47] <cradek> the tool looks awfully big
[14:01:56] <mozmck> yes, that and the measurement text.
[14:02:17] <cradek> what's the tool diameter? or is this the default cone?
[14:02:34] <mozmck> I guess it is the default cone
[14:03:34] <mozmck> If I load a part that is about 14" x 8", the tool and text are small - just about right.
[14:04:54] <cradek> http://timeguy.com/cradek-files/emc/48x96wfm.png
[14:06:00] <mozmck> That looks more reasonable - still fairly large if it increases as you zoom in.
[14:06:36] <cradek> yes - might be better if the numbers would scale if you zoom, but that's not how it works currently
[14:06:47] <cradek> anyway, I can't reproduce your behavior
[14:07:04] <mozmck> axis doesn't use gremlin does it?
[14:07:11] <cradek> yes it does
[14:07:38] <jepler> they share the bulk of their drawing code
[14:07:39] <cradek> gremlin is the preview part of AXIS that was extracted and generalized
[14:07:49] <jepler> right
[14:08:02] <mozmck> If it was extracted, could it have diverged?
[14:08:04] <jepler> but gremlin is also used to refer to the gtk preview widget, making it cnofusing
[14:08:33] <mozmck> Hmm, yes, there is a hal_gremlin widget that I'm using...
[14:08:35] <jepler> if it happens to be happening in a part that is copied, not shared, yes
[14:09:03] <jepler> however, it's my recollection is that the code that draws the dimensions would be shared
[14:10:34] <mozmck> The dimension numbers scale with the part here. They start out quite large and get larger as you zoom in. It would be nice if they started out a readable size and stay that way when you zoom - but that might not be trivial.
[14:12:45] <cradek> emc/task/emctask.cc 391: interp_error: A file is already open
[14:12:45] <cradek> A file is already open
[14:12:45] <cradek> can't open /home/cradek/big.ngc
[14:12:57] <cradek> in master I keep getting these when I reload, and sometimes also on the first open
[14:15:06] <cradek> sim/axis, F1 F2 home o ~/file.ngc return [file loads] Control-r [get these two errors]
[14:18:21] <jepler> cradek: hm thought we'd fixed that one
[14:20:15] <mozmck> It would be nice if there was a way to close a file too
[14:20:45] <jepler> tormach mentioned having a patch to add that; you could get it from them and adapt it to linuxcnc if you feel that way.
[14:21:54] <mozmck> Oh, I might have to try that. I don't open a file to start with, and we had requests to close the open file so I just made a blank.ngc with nothing but M2 in it
[14:21:58] <mozmck> and I load that.
[14:22:29] <mozmck> who would I contact at tormach?
[14:23:16] <jepler> (e-mail in private message)
[17:33:07] <jepler> Subject: [ANNOUNCE] 4.1.27-rt30
[17:33:16] <jepler> (that's about the newest bunch of patches in this latest release)
[17:48:25] <alex4nder> hmm
[20:30:30] <KGB-linuxcnc> 03Gergely Nagymate 052.7 65b9c13 06linuxcnc 10src/hal/drivers/gm.h 10src/hal/drivers/hal_gm.c Fixed some booting issue with GM6-PCI driver * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=65b9c13
[20:30:31] <KGB-linuxcnc> 03Sebastian Kuzminsky 052.7 73918eb 06linuxcnc 10src/emc/task/emctaskmain.cc Merge remote-tracking branch 'origin/2.6' into 2.7 * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=73918eb
[20:31:46] <KGB-linuxcnc> 03Sebastian Kuzminsky 05master 7a8021e 06linuxcnc 10src/emc/task/emctaskmain.cc 10src/emc/usr_intf/axis/scripts/axis.py 10src/hal/components/carousel.comp Merge remote-tracking branch 'origin/2.7' * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=7a8021e
[20:32:53] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15SebKuzminsky commented on issue #83: Merged into 2.7 and master. Thanks for your contribution to LinuxCNC! 02https://github.com/LinuxCNC/linuxcnc/pull/83#issuecomment-228498246
[20:39:09] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15cradek closed pull request #83: Fixed some booting issue with GM6-PCI driver (062.7...06general-mechatronics-PCI-driver-fix) 02https://github.com/LinuxCNC/linuxcnc/pull/83
[21:46:44] <KGB-linuxcnc> 03Jeff Epler 05master 2d68a66 06linuxcnc 10tests/interp/compile/test.sh tests: Get runtime linker errors early * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=2d68a66
[21:46:45] <KGB-linuxcnc> 03Jeff Epler 05master 5e5ae81 06linuxcnc 10tests/interp/compile/use-rs274.cc use-rs274: must initialize variable file name * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=5e5ae81
[21:46:45] <KGB-linuxcnc> 03Jeff Epler 05master fe7b155 06linuxcnc 10tests/interp/compile/use-rs274.cc use-rs274: test that NURBS can be used * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=fe7b155
[21:53:08] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15jepler commented on issue #33: I believe you are correct. I can see this by searching our source tree.... 02https://github.com/LinuxCNC/linuxcnc/pull/33#issuecomment-228502507