#emc-devel | Logs for 2010-07-13

[00:11:11] <cradek> http://sourceforge.net/tracker/?func=detail&aid=2561963&group_id=6744&atid=106744
[00:11:21] <cradek> oops
[01:02:55] <cradek> I tested it on the program that bit me yesterday (starting at TnM6 when lines before and after it called subs; subs defined at the beginning of the file)
[01:03:13] <cradek> so I conclude it works, at least in the easy cases
[05:40:41] <CIA-2> EMC: 03cmorley 07v2.4_branch * r6cd3fab309e3 10/src/emc/usr_intf/pncconf/pncconf.py: add missing spindle-at-speed logic
[05:40:52] <CIA-2> EMC: 03cmorley 07v2.4_branch * r0258c4ca5e58 10/src/emc/usr_intf/pncconf/pncconf.py: Add velocity and acceration restraints to spindle stepper
[14:21:29] <jepler> give Kostas Papadimakis a cookie, he found a bug.
[14:22:42] <skunkworks> heh
[14:23:46] <jepler> I think it's due to a typo that's as old as dirt in some code I incorporated in axis to implement the jog increment combobox:
[14:23:49] <jepler> upvar ::combobox::${w}::oldFocus oldFocus
[14:23:51] <jepler> - upvar ::combobox::${w}::oldFocus oldGrab
[14:23:54] <jepler> + upvar ::combobox::${w}::oldGrab oldGrab
[14:24:14] <jepler> (that's showing what I hope is the fix...)
[14:34:56] <jepler> cradek: is there an SF report for the RFL bug you fixed?
[14:39:10] <cradek> I closed it
[14:39:28] <cradek> http://sourceforge.net/tracker/?func=detail&aid=2561963&group_id=6744&atid=106744
[14:40:23] <jepler> thanks
[14:40:54] <jepler> here's the version I'm looking at putting in 2.4: http://emergent.unpy.net/files/sandbox/0001-Fix-RFL-between-a-sub-and-its-call-SF-2561963.patch
[14:41:19] <jepler> I reworded the change subject, and put both changes into a single commit
[14:41:38] <cradek> looks right to me
[14:41:56] <cradek> (can't believe I screwed that up)
[14:42:17] <jepler> luckily for you, the project doesn't have a rubber weevil to award
[14:42:43] <cradek> surely the weevil would acknowledge my heroism for touching this code and give me a break
[14:44:05] <cradek> (did you test it too? I haven't heard back from micges)
[14:44:46] <jepler> cradek: I don't know how to tell if a weevil acknowledges heroism. I assume it'll just keep staring at you with those faceted eyes in any case
[14:44:59] <jepler> ooh nice photo on wikipedia http://upload.wikimedia.org/wikipedia/commons/c/c4/Long_nosed_weevil_edit.jpg
[14:45:55] <jepler> cradek: no, I haven't tested it yet
[14:46:07] <jepler> I just want to believe it's not wrong
[14:49:00] <jepler> it also wouldn't hurt to have testing on the patch I just attached to https://sourceforge.net/tracker/?func=detail&aid=3028917&group_id=6744&atid=106744
[14:50:16] <e-jones> hi all
[14:50:24] <jepler> hi e-jones
[14:51:44] <e-jones> i would like to ask if there is something like roadmap for emc2?
[14:53:44] <jepler> not really. each developer works on whatever he's interested in and on his own schedule.
[14:54:16] <e-jones> so, works somebody on emc + preempt_rt? :)
[14:54:34] <jepler> there's no current work that I know of
[14:59:55] <jepler> this work was done in 2009: http://mid.gmane.org/200910110020.00011.mb@bu3sch.de
[14:59:59] <SWPadnos> hmmm. does/should the RFL code refuse to start within a subroutine?
[15:00:23] <jepler> however, his approach means that all hardware drivers that use linux module APIs for hardware access (i.e., all of them!) would have to be ported
[15:00:35] <jepler> SWPadnos: no, it'll do something blindingly stupid instead
[15:00:35] <SWPadnos> I think it should, since there is no place to return to when ENDSUB is reached
[15:00:38] <SWPadnos> ok
[15:03:17] <jepler> also the description of his latency numbers ('huge latency spikes', 'jitter can be brought into the millisecond range') I'm not too stoked about preempt_rt performance :-/
[15:05:06] <jepler> finally, I'd argue against including the patch as-is because it copies too much code from sim; before inclusion, a way should be found to share the code. (possibly by using the new 'linux' pthread-style code and degrading gracefully when not running setuid root)
[15:08:00] <cradek> SWPadnos: it appears to do a relatively sane thing (run from there, and then continue after the first call to the sub)
[15:09:01] <SWPadnos> ok, so it doesn't continue with the code after the sub (as though the Oxx SUB hadn't been there)
[15:09:04] <e-jones> jepler: thanks for notes
[15:09:34] <cradek> SWPadnos: no, I was wrong, it does a silly thing as jepler said
[15:10:14] <SWPadnos> heh, ok :)
[15:10:29] <cradek> I think it starts running right after the endsub
[15:10:47] <cradek> (the first time it encounters a runnable line that's >= the selected start line)
[15:11:24] <SWPadnos> that's what I would have expected, with the slight possibility of it stopping with an error (ENDSUB with no call or soemthing)
[15:11:26] <SWPadnos> something
[15:12:06] <cradek> not that, because the program is actually evaluated from the beginning, so the sub/endsub is perfectly fine
[15:12:10] <SWPadnos> oh wait, you're saying that it appears to start execution after the endsub, not that it starts inside the sub and then continues linearly through the endsub?
[15:12:16] <cradek> right
[15:12:18] <SWPadnos> ah, ok
[15:12:26] <SWPadnos> that's not as insane as it could be
[15:12:33] <cradek> not a high bar
[15:12:39] <SWPadnos> nope
[15:12:50] <SWPadnos> at least it's not a game of limbo
[16:38:11] <skunkworks> seb!
[16:38:28] <seb_kuzminsky> hey :-)
[16:39:12] <cradek> seb_kuzminsky: it looks fine to me this morning. I will try from home again later. The problem was the compiler output (some function ``whatever()'' defined but not used) etc
[16:39:23] <cradek> I could have sworn it showed up wrongly on your machine at workshop too
[16:40:09] <seb_kuzminsky> hmm
[16:40:14] <seb_kuzminsky> * seb_kuzminsky looks
[16:40:36] <cradek> I helpfully made a big failed compile yesterday
[16:41:42] <seb_kuzminsky> oh yeah, thanks for that: http://emc2-buildbot.colorado.edu/buildbot-admin/builders/lenny-x86-trunkish-realtime-rip/builds/1173/steps/compile/logs/stdio
[16:41:44] <seb_kuzminsky> that does look bad
[16:42:30] <seb_kuzminsky> oops, i shouldnt paste buildbot admin urls, because they dont work for people...
[16:42:33] <cradek> today it looks fine here
[16:42:38] <cradek> (I removed -admin)
[16:42:54] <seb_kuzminsky> urgh
[16:42:54] <cradek> I can't imagine what I changed to make it look right, but it's fine
[16:43:53] <cradek> view / character encoding / unicode
[16:44:58] <cradek> maybe you can specify the encoding in the html?
[16:45:14] <cradek> (or use LANG=C and make it irrelevant)
[19:17:55] <CIA-2> EMC: 03jepler 07v2.4_branch * rf2668616a6fb 10/share/axis/tcl/combobox.tcl: axis: fix jogs after selecting increment (SF#3028917)
[19:17:57] <CIA-2> EMC: 03jepler 07v2.4_branch * rdd4b81cf8c1b 10/share/axis/tcl/combobox.tcl: axis: fix keys while choosing jog increment (SF#3028197)
[19:18:04] <CIA-2> EMC: 03jepler 07v2.4_branch * r408e928a2ac6 10/share/axis/tcl/axis.tcl: axis: fix high CPU usage on icewm