#emc-devel | Logs for 2009-02-06

[00:00:41] <CIA-2> EMC: 03cmorley 07TRUNK * 10emc2/src/hal/classicladder/spy_vars_gtk.c: oops forgot to remove debug printf...
[00:06:48] <PCW_> PCW_ is now known as PCW
[00:33:44] <steves_logging>
[01:27:58] <cradek> I have massaged this nurbs/spline patch (LOTS of changes to the interp since then). unfortunately there is a file missing; it contained two functions: nurbs_point and alpha_finder.
[01:28:27] <cradek> oops, and knot_vector_creator
[01:30:09] <SWPadnos> perhaps an email is in order
[01:31:23] <cradek> sounds like EBo already has emailed him
[03:15:31] <seb_kuzminsky> poor donnie
[03:15:58] <seb_kuzminsky> he's so confused, and he doesnt even know it
[03:23:06] <seb_kuzminsky> ferror on amp enable? sounds like a misdiagnose to me
[03:23:21] <seb_kuzminsky> ferror only happens if emc2 tells the machine to move, and the machine doesnt, right?
[03:23:35] <seb_kuzminsky> or rather, if the machine strays too far from the commanded position
[03:24:38] <cradek> consider something (the amp) causing motion when emc doesn't ask for it
[03:24:54] <cradek> turn amps on, axis jumps, following error, amps turn off
[03:25:00] <seb_kuzminsky> hmm
[03:25:17] <seb_kuzminsky> like the amps glitch the motor power lines on enable?
[03:25:51] <cradek> or it's miswired, or amp way out of balance, or just about anything else
[03:25:59] <seb_kuzminsky> * seb_kuzminsky sighs
[03:26:32] <cradek> encoder hooked up backward will cause a runaway and immediate ferror
[03:26:44] <cradek> same for tach, if it has them
[03:27:14] <cradek> or just grossly out of tune
[03:27:28] <seb_kuzminsky> should i tell him "bust out halscope"?
[03:27:30] <cradek> next step is troubleshooting :-/
[03:27:43] <cradek> I have no idea what advice would best serve him
[03:27:47] <SWPadnos> he shouldn't be using EMC yet
[03:28:07] <seb_kuzminsky> it seems to me hm2 is kind of a noob magnet
[03:28:07] <SWPadnos> at this stage, it's best to use a bare HAL to get units right
[03:28:35] <SWPadnos> or even voltmeters and batteries and stuff. if the amps fault out immediately, that could be almost anything
[03:28:57] <seb_kuzminsky> he claims if he sets pwmgen scale to 5.5 it seeks tto the home switch successfully
[03:29:09] <seb_kuzminsky> so that supports cradek's out-of-tune theory i think
[03:29:13] <SWPadnos> 4.166 should be the number, I bet
[03:29:35] <SWPadnos> if his drives make 250 IPM at full input command
[03:29:50] <cradek> yeah if it moves in any controlled fashion, it's likely wired right
[03:30:03] <SWPadnos> that's true
[03:30:19] <SWPadnos> unfortunately, I don't know how much of this stuff is in the manuals/wiki
[03:30:28] <SWPadnos> or which ones have it, for that matter
[03:51:45] <cradek> I don't believe in this advice to tune the pid without using emc
[03:51:55] <seb_kuzminsky> oops
[03:52:03] <seb_kuzminsky> how do you think it should be done?
[03:52:44] <cradek> I like to make a gcode loop that moves back and forth with delays, then set up halscope to trigger on the move
[03:53:19] <seb_kuzminsky> when i tuned my servos i did that move, pause, move thing in bash ;-)
[03:53:35] <cradek> I think something like that is gratuitously hard to do without emc running
[03:55:06] <SWPadnos> especially for something like hostmot2, where the lines to load it are much more complex than "loadrt pwmgen"
[03:55:52] <cradek> you can safely start emc with a low velocity and a very weak tuning (like P=1, everything else 0)
[03:55:53] <seb_kuzminsky> certainly use a hal file to do the basic (non-tune) setup, no argument there
[03:56:08] <cradek> then carefully raise P and get it to move
[03:56:13] <seb_kuzminsky> cradek: and high ferror limits i guess?
[03:56:27] <SWPadnos> EMC2 velocity limits have no meaning until the pwmgen scale is set
[03:56:41] <cradek> higher than normal, but low enough to stop if it runs away or oscillates bad
[03:56:50] <cradek> that's another (huge) benefit to having emc running
[03:56:54] <seb_kuzminsky> cradek and jmk should have an edit-war of the Tuning wiki page, then fight about it on a public mailing list
[03:56:57] <cradek> if it goes nuts, it'll just stop
[03:57:13] <seb_kuzminsky> that's a really good point
[03:57:21] <cradek> ask me how I know this!
[03:57:22] <SWPadnos> I was going to suggest loading EMC and then doing some low level stuff with halcmd, but it's a PITA to explain
[03:57:23] <seb_kuzminsky> i had some serious scares when it started oscillating...
[03:57:41] <cradek> you do something wrong, and it goes ZIP about a half inch and stops
[03:58:02] <seb_kuzminsky> i read the story of you guys getting stuart's big machine to oscillate! sketchy!
[03:58:23] <cradek> ack, yeah
[03:58:43] <cradek> it was only a very minor oscillation (with respect to the axis travel, heh)
[03:58:52] <seb_kuzminsky> heh
[03:59:53] <SWPadnos> <1% :)
[04:00:03] <cradek> an inch maybe, a few Hz
[04:00:12] <SWPadnos> ~1/2%
[04:00:18] <seb_kuzminsky> yeah, but in amperes?
[04:00:18] <cradek> very heavy
[04:00:39] <SWPadnos> phooey, who needs those?
[04:00:42] <cradek> hundreds?
[04:01:24] <seb_kuzminsky> i'm looking at the new MillSetup page
[04:01:25] <seb_kuzminsky> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?MillSetup
[04:01:41] <seb_kuzminsky> down where he talks about coordinate systems it's a bit confusing
[04:01:45] <cradek> sorry, they were 45 amp I guess
[04:01:45] <seb_kuzminsky> it says "System Command"
[04:01:50] <seb_kuzminsky> P1 G54
[04:01:52] <seb_kuzminsky> P2 G55
[04:01:54] <seb_kuzminsky> etc
[04:02:12] <seb_kuzminsky> the G-code command you enter is G54 or G55 etc
[04:02:20] <seb_kuzminsky> and you end up in coord system 1 or 2 etc
[04:02:21] <seb_kuzminsky> right?
[04:02:23] <cradek> it's a two column table, poorly typeset
[04:02:32] <cradek> yes
[04:02:42] <seb_kuzminsky> ok i'll clean that up
[04:02:59] <cradek> G54 puts you in system 1, which is set with G10L2 P1
[04:03:12] <cradek> G55 ~~ system 2 ~~ G10L2 P2
[04:08:10] <cradek> note to self: need to fix diameter mode's interaction with touch off (ugh)
[04:14:48] <seb_kuzminsky> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?action=history&id=MillSetup
[04:17:20] <cradek> slick
[04:18:01] <seb_kuzminsky> wow, i'd have settled for: <cradek>: correct
[04:18:04] <seb_kuzminsky> :-)
[04:18:52] <cradek> it's neat how several people have jumped in to work on the new page already
[04:19:31] <seb_kuzminsky> yeah, emc2's got a pretty good community
[04:25:35] <cradek> [2760022.174186] 10719916: ERROR: Exceeded soft limit
[04:25:38] <cradek> from my dmesg
[04:25:43] <cradek> wonder what those odd numbers are
[04:25:51] <cradek> well i know what the [first] one is
[04:26:26] <cradek> ./emc/motion/control.c: reportError("Exceeded soft limit");
[04:27:07] <cradek> ah, emcmotStatus->heartbeat, whatever that is
[04:45:36] <dgarr> a fix for at_pid.c (i think -- works for me, prior version not so much) http://pastebin.ca/1328713
[04:47:12] <cradek> dgarr: thanks! I thought I found all of those erroneous param-to-pin changes but they still keep popping up.
[04:47:42] <dgarr> you're welcome
[04:48:28] <cradek> line 708's change is just stylistic?
[04:49:13] <dgarr> i've turned that machine off for the night -- what number in the diff?
[04:49:22] <cradek> 63-64
[04:49:44] <cradek> the last block
[04:50:01] <dgarr> yes, style to agree with other changes at params to pins
[04:50:43] <cradek> aside from this mess, does autotuning work for you?
[04:51:42] <CIA-2> EMC: 03cradek 07TRUNK * 10emc2/src/hal/components/at_pid.c: patch from dewey garrett: fix more breakage caused by param-to-pin changes
[04:52:03] <dgarr> i used it orignally (about 9 months ago) and it is interesting and maybe a good starting point.
[04:52:25] <cradek> neat.
[04:52:33] <dgarr> i haven't tuned (using at_pid) recently but I always mean to use it some more
[04:53:34] <cradek> I have not seen it work well enough to be much better than nothing, but I always blame it on being the wrong kind of machine, or backlash, etc.
[04:53:45] <cradek> for a tight heavy machine I think it might work
[04:54:45] <cradek> thanks again, and goodnight - I'm off
[04:54:47] <dgarr> the moment of inertia on my rotary axis can change a lot depending on whether i use a heavy chuck or just a taper so tuning can be helpful to get speed up -- however, there are many other things to do before trying get speed up
[04:54:53] <dgarr> gnight
[04:55:06] <seb_kuzminsky> seeya cradek
[10:46:41] <alex_joni> real 1m34.074s
[10:46:41] <alex_joni> user 4m36.460s
[10:46:41] <alex_joni> sys 1m2.160s
[10:46:55] <alex_joni> for a fresh compile of emc2 on this machine
[11:29:11] <CIA-2> EMC: 03alex_joni 07TRUNK * 10emc2/scripts/emc.in: don't run axis-remote on machines without X
[11:33:22] <CIA-2> EMC: 03alex_joni 07TRUNK * 10emc2/src/emc/usr_intf/keystick.cc: fix 2 warnings on gcc 4.3.2
[11:51:00] <CIA-2> EMC: 03alex_joni 07TRUNK * 10emc2/src/hal/utils/halrmt.c: fix 11 warnings on gcc 4.3.2, added a FIXME comment - return values for various functions aren't really used
[12:07:31] <CIA-2> EMC: 03alex_joni 07TRUNK * 10emc2/src/emc/rs274ngc/rs274ngc_pre.cc: fix warnings on gcc 4.3.2
[12:10:25] <alex_joni> morning BigJohnT
[12:12:05] <BigJohnT> morning alex_joni
[12:12:20] <BigJohnT> man my back hurts today
[12:12:37] <BigJohnT> slipped on the ice :/
[12:14:03] <alex_joni> mine does too
[12:14:19] <alex_joni> didn't slip though.. probably picked something up wrognly
[12:16:13] <BigJohnT> my wife has that problem she can just bend over to tie her shoe and wham her back goes out
[12:16:50] <alex_joni> that's what happens today
[12:16:56] <alex_joni> I bent forward to take a picture
[12:17:05] <alex_joni> when I wanted to go back up, it didn't work :/
[12:17:56] <BigJohnT> the doctor told her it was from lack of use of those muscles
[12:22:58] <alex_joni> I don't think it's a muscle thing for me
[12:24:23] <BigJohnT> bad disk?
[12:28:08] <alex_joni> not sure..
[12:29:16] <alex_joni> does anyone mind if I build sim packages for 8.10 ?
[12:29:30] <BigJohnT> in her case it was not the muscle but the silverskin as we call it here between the muscles
[12:29:46] <BigJohnT> I don't mind :)
[12:36:03] <CIA-2> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/quickstart/stepper_quickstart.lyx: change page numers to left/right
[12:46:28] <CIA-2> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/motion/ (kinematics.lyx tweaking_steppers.lyx): fix links that point to other manuals
[12:56:36] <jepler_> jepler_ is now known as jepler
[12:58:02] <CIA-2> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/install/Latency_Test.lyx: change page numers to left/right
[13:19:51] <BigJohnT> * BigJohnT hobbles off to work
[13:41:05] <jepler> good morning
[13:41:48] <alex_joni> hi
[13:43:07] <micges> good morning
[13:43:40] <micges> I have problem and don't know it is error or no
[13:43:46] <micges> I have 8 tools
[13:44:21] <micges> question is:
[13:44:30] <micges> when I T1M6
[13:44:39] <micges> G43 H1
[13:44:51] <micges> then I touch off Z axis
[13:45:28] <micges> then MDI G0Z100
[13:45:32] <micges> then MDI T2M6 G43 H2
[13:45:41] <micges> then I MDI G0Z0
[13:46:33] <micges> z axis MUST go to the same position when was touch off ?
[13:48:42] <cradek> if your tool table is right, it will
[13:49:22] <jepler> that's what it did for me -- sim/axis, I have tool 1 with length .5 and tool 2 with length .7. I worked in the "X" view, and after I typed the final G0Z0 in MDI the bottom of the tool is touching the red "Y" line that is horizontal on the screen
[13:50:46] <micges> ok next: I have program loaded after first touch off
[13:50:49] <alex_joni> dpkg-buildpackage -j4 is fun
[13:51:01] <micges> at level of material i Z0
[13:51:23] <micges> I'm run program and break in middle
[13:52:05] <micges> move Z up
[13:52:24] <micges> change tool and apply tool offset of that tool
[13:52:26] <alex_joni> break.. you mean abort?
[13:52:37] <micges> abort yes
[13:52:40] <jepler> one thing to keep in mind when aborting during the middle of the program is that the interpreter's state is almost always from some place further down the program, maybe all the way to the end.
[13:53:26] <jepler> but, go on
[13:54:51] <micges> and restart at aborted line then Z position must be the same as in Gcode ?
[14:37:47] <micges> when running program from begin to end everything works fine, when breaking and changing tool and continuing didn't work
[14:38:01] <micges> (It mustn't work)
[14:38:54] <micges> bbl
[14:41:15] <cradek> I wish cvsweb would show both the diff and the log message. you can get either the diff, or the log and the full file
[14:42:24] <cradek> oh god, perl
[14:46:24] <jepler> hah!
[14:49:45] <cradek> patch just skips crap before the diff, right?
[14:52:57] <jepler> yes that's my experience
[14:54:47] <jepler> you could try the gitweb interface to emc2 cvs -- it's updated daily now. http://git.unpy.net/view/emc2.git it gives nice diffs with log messages, and works by changesets: http://git.unpy.net/view?p=emc2.git;a=commitdiff_plain;h=c2e0509c70ef157a73e96e7eaf8b89b9a5c1619f
[15:01:47] <cradek> neat
[15:01:57] <cradek> this means I don't have to figure out the perl
[15:03:10] <jepler> heh
[15:03:31] <jepler> a new cvs web interface based around cvsps would be cool
[15:03:42] <jepler> if you could get cvsps incremental mode to work reliably, anyway
[15:04:08] <cradek> but its diffs are bogus: http://git.unpy.net/view?p=emc2.git;a=commitdiff;h=db9872536322220f722a66f1b1a15c92fa372328
[15:04:21] <cradek> the first line of code is smushed up after the @@
[15:04:44] <cradek> and I don't get a fixed font, which makes them hard to read
[15:05:12] <cradek> ok, there is a "plain" which gives a good font, but the diff is still broken
[15:05:27] <jepler> did you check that patch doesn't accept it? I think that what you're looking at is a feature
[15:05:44] <jepler> it's "diff -p"
[15:05:45] <jepler> -p --show-c-function
[15:05:45] <jepler> Show which C function each change is in.
[15:06:05] <jepler> patch should ignore it, but it can be useful to humans
[15:06:06] <cradek> O RLY
[15:06:15] <cradek> I sit corrected
[15:06:53] <jepler> on http://git.unpy.net/view?p=emc2.git;a=commitdiff;h=db9872536322220f722a66f1b1a15c92fa372328 I get a fixed-width font, so it must depend on browser settings
[15:06:59] <cradek> ok, now that I know that, I think I like it
[15:07:49] <jepler> (but I wouldn't try to patch except from the 'plain' page!)
[15:08:12] <cradek> definitely
[15:14:22] <jepler> the reason you're getting variable-width font may be this bug, fixed in ff2.0: https://bugzilla.mozilla.org/show_bug.cgi?id=159440
[15:22:11] <cradek> \todo FIXME documentation problem I couldn't get this equation to render right since doxygen gets confused and interprets a less than symbol as a control character, so I changed it to less than or equal to. does it make any difference?
[15:24:12] <BigJohnT> what happens when you use two less than characters?
[15:25:43] <cradek> I don't care - I was just making fun of doxygen (and the use of doxygen to typeset equations)
[15:26:30] <BigJohnT> oh I missed that it must be the drugs :)
[15:29:08] <cradek> I still don't "get" git. There's the list of "heads" which correspond to our branches. But when I visit one, there is stuff listed that I know wasn't done on that branch
[15:30:04] <jepler> maybe its import of branches is still wron
[15:30:05] <jepler> g
[15:30:24] <cradek> http://git.unpy.net/view?p=emc2.git;a=log;h=refs/heads/rigid_tap
[15:30:45] <cradek> the first 6 (?) changes seem like they could be right
[15:34:34] <cradek> maybe I don't know what "heads" is. it may not know about branches, only that that rigid_tap spot has no children - it's an endpoint in the "tree" of changesets
[15:35:16] <jepler> the ones before that are from TRUNK before the branch was created. http://emergent.unpy.net/files/sandbox/rigid-tap-graphically.png
[15:35:18] <cradek> so, I can't see where development was split on purpose back in the cvs days in order to make rigid-tap related changes
[15:35:47] <cradek> ok I think that's what I was just trying to explain
[15:35:59] <cradek> I understand now
[15:36:00] <jepler> I agree; on that web page you can't tell which is the first revision that is on the branch
[15:36:42] <cradek> but gitk does show that information - it sees the split (I hesitate to say branch)
[15:36:46] <cradek> that's good to know
[15:38:32] <jepler> if you were in a clone of that repo, you could see the revisions you were interested in by "git-log -p master..rigid_tap"
[15:39:27] <jepler> but I can't tell how to get that from gitweb
[15:39:38] <BigJohnT> would adding \usepackage{fancyhdr} in the preamble of the lyx docs cause a problem?
[15:40:09] <jepler> BigJohnT: I don't know
[15:40:53] <BigJohnT> ok, it gives you better control of the pdf page layout...
[15:40:57] <cradek> jepler: it's nice to see that git does a sane cvs import
[15:41:45] <cradek> we probably won't want to be the last ones to stop using cvs
[15:42:04] <seb_kuzminsky> i think we already are cradek :-P
[15:42:12] <jepler> BigJohnT: if \usepackage{fancyhdr} works on both dapper and hardy, then it's OK to introduce in TRUNK for some benefit.
[15:42:34] <BigJohnT> ok, I'll try it at home on both first
[15:42:38] <cradek> seb_kuzminsky: I wouldn't be surprised if cvs was still one of the two most commonly used systems (svn being the other, for some reason)
[15:42:46] <cradek> (wonder if google knows)
[15:43:30] <jepler> BigJohnT: but if it requires an additional package to be installed, then it's not OK for 2.2
[15:44:27] <BigJohnT> ok, I was just going to try it on 2.3
[15:44:30] <BigJohnT> thanks
[15:45:04] <seb_kuzminsky> jepler: i'm curious why even build-deps are off-limits in 2.3?
[15:45:18] <jepler> seb_kuzminsky: that's not what I meant to say
[15:45:32] <jepler> changing 2.2 build deps is bad, changing TRUNK build deps is OK until 2.3.0.
[15:45:48] <seb_kuzminsky> oh yeah, that *is* what you said
[15:45:55] <seb_kuzminsky> * seb_kuzminsky look for his coffee
[15:45:58] <jepler> heh
[15:47:04] <jepler> and by that I mean "making a change that creates new real dependencies"; if build-depends are inaccurate, then they will be fixed no matter what stage of development.
[15:47:36] <seb_kuzminsky> sure
[15:48:05] <seb_kuzminsky> i have a half-finished hardy non-rt vm somewhere that'll run a pbuilder buildslave, that should ferret out any broken build-deps
[15:48:22] <jepler> BigJohnT: I don't remember if, when you wanted to introduce \usepackage{tocbibind} if me building on an even older than dapper system was an issue, but it isn't now -- the website docs are built on hardy.
[15:48:54] <jepler> I looked for the irc logs but got bored before I found the important conversation :-P
[15:51:28] <BigJohnT> ok I think that I got the toc issue to go away without it since then... hmm, don't even remember the exact reason I wanted to use it
[15:51:49] <jepler> heh
[15:52:10] <jepler> it was like a year ago :-P
[15:52:36] <BigJohnT> yea, and I've slept since then :)
[15:53:04] <seb_kuzminsky> is there a search function for the irc logs or do you wget them all and grep?
[15:53:19] <cradek> google does a decent job
[15:53:22] <jepler> seb_kuzminsky: google
[15:53:27] <seb_kuzminsky> right-o
[15:54:56] <jepler> once in awhile I do use my shell access on the webserver to grep around, particularly for recent logs
[18:02:12] <Guest575> Guest575 is now known as skunkworks