#emc | Logs for 2009-02-27

[03:12:18] <dgarr> some more gdb backtrace output from halscope segfault: http://pastebin.ca/1348450 http://pastebin.ca/1348451
[03:12:18] <dgarr> apparently some magic with installing libgtk2.0-0-dbg, not sure it is any more help however
[03:12:18] <dgarr> maybe i need another dbg package for other libs
[03:12:34] <cradek> cool
[03:14:00] <cradek> hm, both those look like they are not necessarily halscope bugs
[03:14:26] <cradek> it's all widget drawing in response to an expose
[03:15:01] <cradek> looks like this is the same gtk version I have on dapper
[03:15:50] <cradek> slow X display still helps you get this?
[03:16:03] <cradek> (if so, that's another reason to believe it's a gtk problem)
[03:16:26] <dgarr> no,this was from this afternoon, display on the computer with emc running
[03:17:46] <dgarr> the segfaults occur more frequently when application and display are on the same machine, eg, harder to reproduce with remote display
[03:17:59] <cradek> oh ok, I thought it was the opposite
[03:18:09] <cradek> have you asked google about this backtrace?
[03:18:39] <dgarr> just a little, most of it is over my head
[03:19:02] <cradek> wonder why I'm not seeing this.
[03:19:57] <dgarr> maybe i should do a whole new checkout (?)
[03:36:22] <jepler> anything specific you do to trigger it?
[03:38:25] <dgarr> no -- sometimes a button push, sometimes moving the cursor across widgits, i have seen it once when the signal that was assigned to trigger changed to trigger it.
[03:39:34] <cradek> I'm no longer convinced I know that this happened to me or anyone else but you
[03:39:54] <dgarr> would you say a fresh checkout is in order then?
[03:40:06] <cradek> maybe we should treat it like a problem with your system - run memtest, clean/rebuild emc, ...?
[03:40:49] <cradek> but you surely run other gtk apps, right?
[03:43:14] <dgarr> the machine where i see these segfaults is in the shop -- i run gimp, inkscape, firefox, maybe some others on occasion, none of these have faulted. i have only seen halscope fault when running RIP built from trunk
[03:44:36] <cradek> ok
[03:47:36] <dgarr> i'm going to try a fresh checkout from cvs
[03:49:43] <jepler> the traceback doesn't mean much to me
[03:49:59] <jepler> I can get some diagnostics from valgrind using roll mode, but it sounds like that's unconnected
[04:05:54] <jepler> 'night folks
[04:30:18] <dgarr> so i built RIP from a fresh trunk checkout: an initial halscope sigsegv, but then none after restarting: http://pastebin.ca/1348488
[04:47:47] <ibuffy> http://blender-archi.tuxfamily.org/Features_requested_for_Architecture
[04:49:00] <ibuffy> Could Blender ever be used for precision machining?
[04:49:08] <ibuffy> Here's a little discussion on the topic: http://projects.blender.org/forum/forum.php?thread_id=50579&forum_id=174
[04:49:24] <ibuffy> But I was hoping you all could chime in
[04:49:26] <cradek> dgarr: are you using the default gtk theme?
[04:49:44] <cradek> also, it is dapper, right?
[04:49:47] <ibuffy> What's the difference between what architects require and what machinists require?
[04:53:04] <dgarr> my theme is the default i think -- i changed colors, no image on background, os is hardy
[04:53:34] <ibuffy> another article on the topic, 2.5d cnc machining with blender: http://www.wikkidwidgets.com/CNCMachiningwithBlender%20-%202.5D.HTML
[04:54:00] <cradek> dgarr: interesting, I was testing on dapper. that's a major difference.
[04:55:23] <dgarr> i should have mentioned that earlier i guess
[04:55:27] <cradek> np
[04:55:43] <cradek> I'll try to find time to play on my hardy machine this weekend
[04:55:53] <cradek> but for now, goodnight
[04:56:04] <dgarr> gnight -- thanks again
[04:56:06] <cradek> thanks for your good reports
[06:20:54] <ibuffy> has anyone given thought to my questions two hours ago?
[06:45:28] <KimK> ibuffy: on architects vs. machinists?
[06:46:22] <ibuffy> KimK: yes
[06:46:57] <ibuffy> And is it possible that Blender can be a CAD program for precision machining
[06:47:24] <ibuffy> not with its current code of course...
[06:51:49] <ibuffy> is it a stupid wish?
[06:51:53] <ibuffy> why, if so?
[06:52:13] <ibuffy> what are the current open source CAD programs lacking?
[06:52:26] <ibuffy> ..compared to..?
[06:56:06] <KimK> I'm not an expert on the latest CAD programs. I'm currently using Qcad because I've switched to Ubuntu, and Qcad is either standard or available with a click using Synaptic.
[06:57:26] <KimK> But I miss TurboCAD Professional that I used to have in Windows. The "Professional" version is 3D. Also has a built in photo renderer, nice for visualizing things.
[06:58:19] <KimK> And TurboCAD was affordable for the average person. Now it lists at $999, I think, but it is frequently discounted to $249 or so.
[06:59:13] <ibuffy> right, qcad doesn't support 3d, not even close to what i'm looking for.
[06:59:49] <KimK> As far as a new open source 3D CAD program, I'm all for it, we need it. And Blender could be the renderer, sure. Isn't there also POVray (spelling?)
[07:00:09] <ibuffy> i didn't give qcad much of a glance
[07:00:13] <KimK> Maybe it's POV-ray
[07:00:23] <ibuffy> yeah, pov-ray
[07:00:39] <ibuffy> opencascade is a 3d CAD program...
[07:00:55] <ibuffy> but i don't think it's very far along..
[07:00:59] <ibuffy> it's why i come in here..
[07:01:19] <KimK> Qcad isn't bad for what it is, a 2D non-parametric program. It's a little like TurboCAD. I think I've heard AutoCAD users say it's a little like AutoCAD too.
[07:01:19] <ibuffy> you professionals, chime in please. i pay well (not really)
[07:02:21] <KimK> Once you get on to the Qcad way of doing things, it's not bad for 2D. Maybe I could find something I can post, let me look around.
[07:12:15] <KimK> Here's a Qcad screenshot of a fixture plate (universal?) with a bunch of tapped holes for putting blocks at convenient places. http://imagebin.ca/view/Ngvzy8.html
[07:13:35] <KimK> Here all the layers are on at once, but if you have a drawing and Qcad, you can turn the layers on and off individually to clarify things.
[07:14:37] <KimK> I didn't put in any dimensions, because I just wanted to get it drawn and either export it or pick up by hand the points I needed for CNC.
[07:15:04] <KimK> CNC, or DRO in this case, as it turned out.
[07:15:49] <KimK> Let me see if I can find anything better...
[07:21:26] <KimK> Here's a prelim sketch for an MPG knob. http://imagebin.ca/view/JbA1JmH6.html
[07:22:35] <KimK> You might want to click on the image to see it full size to get rid of the optical artifacts. Anyway, it shows that Qcad does a pretty good job on laying out rotary angles.
[07:25:25] <KimK> But yes, you're right, Qcad leaves much to be desired.
[07:26:59] <KimK> * KimK will be back after a snack, keep chatting, and I'll read it in a half-hour or so.
[09:28:37] <KimK> * KimK came back after a break, but saw no chat
[09:31:34] <KimK> I'll check back later
[10:54:41] <JymmmEMC> http://woodgears.ca/marbleadd/index.html
[13:51:25] <Kohlswa> yay!! sometimes its good to arrive early to3 work, got myself a gigantic laser printer.
[14:47:46] <twice2> hello, i'm trying to understand why the keystick ui keeps crashing to the diagnostic screen
[14:55:18] <jepler> I don't think any of the developers use keystick beyond seeing whether it starts at all (and that not very often)
[14:55:39] <jepler> if you isolate and fix any problems with it, please submit a patch
[15:14:00] <twice2> jepler:ok, will do, glad to see you
[15:21:30] <jepler> if you can figure out a way to make it crash consistently, that'd probably get a developer interested enough to look into it as well
[15:25:42] <twice2> jepler, i can crash it by simply cycling windows with the <?> key
[15:27:17] <cradek> it doesn't crash when I do that
[15:27:38] <jepler> after I held down ? for a few seconds mine has freaked out and I have a blank display
[15:28:01] <cradek> trunk?
[15:28:15] <jepler> yes
[15:28:19] <cradek> huh
[15:30:34] <jepler> ==16488== Process terminating with default action of signal 3 (SIGQUIT)
[15:30:34] <jepler> ==16488== at 0x4000772: (within /lib/ld-2.3.6.so)
[15:30:34] <jepler> ==16488== by 0x40AAFE9: _nc_outch (in /lib/libncurses.so.5.5)
[15:31:11] <jepler> hmmm it doesn't hang when not running under valgrind!
[15:31:26] <cradek> ugh.
[15:31:36] <jepler> must be dependent on system speed or something awful like that
[15:32:04] <cradek> maybe twice2 could attach the debugger and than crash it
[15:32:43] <cradek> twice2: what version are you running?
[15:34:57] <twice2> cradek: version 2.2.8, yes do docs have info on using the debugger?
[15:35:26] <cradek> twice2: start emc, pull up another terminal
[15:35:32] <cradek> at the new terminal
[15:35:38] <cradek> gdb keystick `pidof keystick`
[15:35:45] <cradek> at the (gdb) prompt, type continue
[15:35:54] <cradek> now cause keystick to crash
[15:40:05] <twice2> cradek:ok, crashed in gdb, does gdb dump something somewhere?
[15:40:21] <cradek> at (gdb), type bt
[15:40:31] <cradek> you'll get a backtrace
[15:40:35] <cradek> put that on pastebin please
[15:41:41] <twice2> i think i need to terminate emc first?
[15:41:59] <cradek> no
[15:42:14] <cradek> it should stop right at the crash
[15:42:21] <cradek> then you ask for the bt
[15:42:57] <jepler> if the problem you're seeing is that keystick becomes unresponsive (as opposed to closing its window and exiting), then hit ctrl-c in the terminal with gdb, then type "bt" at the prompt.
[15:43:18] <twice2> ok, think i have something for pastebin
[15:43:43] <cradek> I assumed "crashed in gdb" meant you got an error in gdb, and a gdb prompt
[15:44:35] <twice2> http://pastebin.ca/1348789
[15:45:54] <cradek> I don't think that's the crash
[15:46:02] <cradek> can you describe exactly what you see when it 'crashes'?
[15:46:51] <twice2> cradek: brb
[15:46:57] <JymmmEMC> How many sq miles in a 2 mile radius? = Pi*2^2 = 12.566 Is Pi*R^2 the correct formula?
[15:47:28] <cradek> dude, ask the internet
[15:47:37] <cradek> 'area of a circle'
[15:47:41] <JymmmEMC> I did and got three answers.
[15:48:01] <cradek> three different answers??
[15:48:17] <cradek> use the first hit :-)
[15:48:23] <JymmmEMC> lol
[15:49:08] <JymmmEMC> cradek: "Hey, why didn't I think of that!" =)
[15:49:53] <cradek> but yeah, pi * radius^2 is your formula
[15:50:10] <cradek> it would be very puzzling if you found another formula...
[15:50:43] <JymmmEMC> R2*Pi
[15:51:35] <twice2> cradek: ok, gdb hooked keystick pid, when i crashed keystick no errors showed up in gdb. bt did nothing until i tried to close keystick
[15:52:03] <cradek> ok, thanks
[15:52:34] <jepler> when you say thay keystick "crashes", you mean that its window stays open but it is no longer responsive, correct?
[15:54:05] <twice2> mine becomes unresponsive on the diagnostics screen, but two of the heartbeats are incrementing
[17:19:25] <jepler> twice2: I found and fixed a problem in keystick that might be related to what you saw. I put the fix in CVS TRUNK only, but when we announce 2.3beta in the next week or so you'll be able to give it a try without compiling anything
[17:19:40] <jepler> on the other hand, if you are familiar with compiling emc2, then you can get the change out of CVS right away
[17:28:12] <twice2> ok thanks, i need to get uts on compiling emc i'm guessing i could look at the keystick src in cvs for your comments
[17:30:37] <jepler> the change I made: http://cvs.linuxcnc.org/cvs/emc2/src/emc/usr_intf/keystick.cc.diff?r1=1.15;r2=1.16
[17:39:11] <twice2> can i assume lines beginning with '-' you removed and with '+' you added to keystick.cc?
[17:39:41] <jepler> yes. that's called a diff. it's automatically produced by the program "diff" and automatically applied to a file by "patch"
[17:40:00] <twice2> ic, thank you
[17:40:03] <jepler> google provides lots of informative looking hits for searches like: using diff and patch
[18:10:15] <motioncontrol> good evening.I have the problem whit custom m fuction.i write m119 and i want pass the float value into m fuction.i have write in m119 code: p=$p echo halcmd setp mul2.0.in0 = $p exit 0.when i write m119 p180 the error in terminal is : value '=' invalid for float because?
[18:34:29] <motioncontrol> good evening.I have the problem whit custom m fuction.i write m119 and i want pass the float value into m fuction.i have write in m119 code: p=$p echo halcmd setp mul2.0.in0 = $p exit 0.when i write m119 p180 the error in terminal is : value '=' invalid for float because?
[18:53:59] <jepler> motioncontrol: because setp doesn't include a "="
[18:54:07] <jepler> setp pin-or-param-name value
[18:54:09] <cradek> several problems: P,Q are passed as command line arguments, so you need to access them as $1 and $2. also you have the setp command line wrong.
[18:54:34] <jepler> and what cradek says is correct too
[18:56:30] <motioncontrol> i want set the mul2.0.in0 at value di p parameter of m119 p180 , whath is the syntax correct ?
[18:57:23] <motioncontrol> halcmd mul2.0.in0 $p ?
[18:58:13] <motioncontrol> halcmd setp mul2.0.in0 $p ?
[18:58:13] <jepler> uh no
[18:58:27] <jepler> the second one, if the value you want is in $p (but it doesn't start there, as cradek told you)
[19:00:32] <motioncontrol> jepler excuse you can paste bin one exaple for pass the float value in m custom code i don't understand.i want write m119 p10 and 10 setp in mul2.0.in0.please
[19:00:54] <jepler> no, I think it would be more constructive for you to learn it for yourself
[19:01:51] <motioncontrol> ok thanks i prove
[19:03:12] <jepler> there are two subproblems here. The first ("How do I refer to the P-number in a script called by M1xx") is answered in the "M101" sample script that comes with emc. The second is exactly the same as something you should already know because you must have used "setp" in your .hal files already
[19:03:29] <jepler> put the solutions to the two subproblems together, and you'll have your solution without it being handed to you on a silver platter
[19:04:03] <jepler> and if you're having trouble with that anyway, the terminal is your friend -- experiment with it
[19:05:15] <motioncontrol> ok thanks for you council
[19:07:08] <motioncontrol> i have start the m101 , and this m fuction write 1.0000 and 1.0000 the p and q value .but p and q value not is exsternal value but is declare internaly at m101 p=S1 and q=$1
[19:08:42] <jepler> when you write "P" instead of "p", "S" instead of "$", and so on I can sometimes guess what you mean, but the shell will not be so forgiving.
[19:10:29] <motioncontrol> excuse my error for editing P=$1 and Q=$1 is correct
[19:36:33] <rob__> hi
[19:37:02] <rob__> what does pid.<loopnum>.maxoutput set? the max analog signal?
[19:37:27] <cradek> the maximum value of pid.0.output
[19:37:32] <cradek> [absolute value]
[19:38:24] <cradek> rob__: http://www.linuxcnc.org/docs/2.2/html/man/man9/pid.9.html
[19:38:38] <cradek> these docs are also on your system: type "man pid" at a shell
[19:40:18] <rob__> yea i read manual but still little fogy what value is caped, i see it is set to max_vel value in ini file
[19:43:45] <cradek> the pid output is capped. what that means depends on the rest of your setup -- if pid output is hooked to a dac with dac scaling 1, pid maxoutput will limit the voltage out of the dac
[19:43:57] <cradek> but if dac scaling isn't 1, obviously it will limit it to some other number
[19:44:37] <cradek> setting it to max velocity might make sense if you have velocity mode servo amps that are carefully scaled, but usually that isn't going to be the case
[19:50:17] <rob__> right makes more sence now
[19:52:17] <rob__> so stops PID system comanding a value grater than the system can handle
[19:53:19] <cradek> yes
[19:53:40] <cradek> more importantly, if your output (whatever it is) is saturated, it doesn't keep winding up the integrator
[19:53:52] <cradek> because that's never what you want to happen...
[19:55:36] <cradek> my lathe's servo amps accept a max of +-8.5 volts. I used pid maxoutput to make sure the dacs never give more than that.
[19:56:16] <rob__> all is clear now how it is/can be used in each setup
[19:56:46] <rob__> thx
[19:56:52] <cradek> welcome
[20:56:10] <rob__> is there a spindle-fault input pin?
[21:02:11] <cradek> what behavior do you want when the spindle faults?
[21:02:36] <cradek> I think I'd put that in the hard estop chain
[21:03:50] <pjm_> evening, i have my inverters fault o/p wired into the e-stop chain, so far its not faulted as far as i know
[21:05:00] <cradek> a vfd fault sure might mean something serious is happening - good to estop IMO
[21:05:18] <pjm_> yeah exactly, thats what I thought
[21:05:31] <pjm_> i looked at the Anders-Wallin site for some ideas
[21:15:46] <Jon_geo01005> Any suggestions of references where I can get my mind around the list_for_each function?
[21:23:18] <rob__> i throught might be nice to have fault input that could tell operator spindle fault, axis fault, but as u say also do an Estop, but at same time operator will know why and know to investigate etc
[21:24:23] <jepler> rob__: you could always use a vcp to display the signal
[21:24:51] <rob__> true could make a fault tab
[21:25:12] <cradek> I agree. there isn't currently that input, and once you get an estop and everything turns off it could be hard to tell what did it. but, if it's like my VFD, it will display the fault code on its little screen
[21:25:56] <jepler> most important is that all fault sources go into the external e-estop chain .. then for the ones that you'd like to give a distinct indicator to, bring them into the PC and use a pyvcp led..
[21:26:38] <rob__> yes i agree
[21:27:38] <pjm_> rob__ did u decide on a spindle load feedback method
[21:27:38] <cradek> I also have a fault going the other way - an estop causes the vfd to fault ("external fault" it says)
[21:27:43] <jepler> you can also use a variant of the 'barn door' component to present it as an operator error. http://emergent.unpy.net/index.cgi-files/sandbox/error.comp
[21:27:55] <rob__> just i saw .amp-fault-in and wunder is tehre a spindle one or such did i miss it in the doc
[21:28:11] <cradek> (it can be hard to get out of estop when everything can stop everything else)
[21:28:22] <cradek> nope, only amp fault
[21:30:24] <rob__> hi pjm_ , not yet but i think ill just use a ADC chip prob one kirk linked to, as i will have a encoder on the spindle at a later date for rigid tappin etc
[21:31:00] <rob__> mainly spindle load i want to feed back in, so ican see how hard im pushing the spindle etc
[21:38:13] <twice2> pjm: nice work on your mill! qrp cw moon bounce, cool beans
[21:39:34] <pjm_> ah cool ;-)
[21:39:57] <pjm_> yeah my mill is progressing still, need to do more work on it, and upgrade to a mesa IO card
[21:40:25] <bbernd> Hallo Kann mir jemand sagen warum diese Zeile in der *. hal nicht funktioniert: net Z-home m5i20.0.in-02 => axis.2.neg-lim-sw-in
[21:40:53] <archivist> ebay toy just up the road for you pjm_ 250379159692 (bit large maybe)
[21:41:36] <twice2> pjm: was that cast al-u-minium, nice chips
[21:42:25] <pjm_> archivist - i think my missus would kill me if i bought that, but yeah its only 25 min drive away
[21:42:33] <pjm_> twice2 which part are u referring to?
[21:45:20] <rob__> u guys in UK to then
[21:45:27] <pjm_> i'm down in Poole
[21:45:42] <rob__> im in FENs
[21:46:02] <pjm_> i think archivist is up near leeds/ mancs
[21:46:04] <pjm_> somewhere up there
[21:46:26] <archivist> sunny Swadlincote east midlands
[21:46:44] <archivist> 60 miles south of leeds
[21:47:05] <twice2> you may know where nebraska is :)
[21:47:17] <archivist> over the pond
[21:47:22] <pjm_> i've heard of that place
[21:48:01] <twice2> what is your lat roughly
[21:48:42] <pjm_> im in IO80
[21:48:42] <rob__> pjm_, this is my work in progress http://cnczone.com/forums/showthread.php?t=73943
[21:49:50] <pjm_> wow that is a monster! a nice machine
[21:50:39] <Dallur1> nice
[21:52:05] <rob__> u wanna see our big lathe, nice 35hp to play with on that things, lets just say it will part of tool holders nicly if u get offset wrong lol
[21:53:22] <pjm_> i'd love to convert a proper machine to EMC rather than my little toy mill
[21:53:39] <pjm_> i've got my eye on a bridgeport interact-1 which will probably be scrapped soon
[21:54:41] <twice2> where in G.B. were all the RSG'S sent to break enigma machine?
[21:54:58] <archivist> Bletchley
[21:55:27] <twice2> ah, sounds familiar
[21:56:21] <archivist> there is a museum there now sowing some of the early code braking and computers
[21:57:24] <twice2> most appropriate
[21:57:33] <pjm_> ah yes i need to drive up and visit that place
[21:59:50] <twice2> call to feed hear, keep the power on guys :)
[22:00:43] <archivist> pjm_, Ive been :)
[22:01:13] <archivist> * archivist wants to make sure he gets an even better look next time
[22:01:42] <pjm_> cool, yeah its on my list
[22:02:14] <archivist> they have one of these http://www.collection.archivist.info/archive/Lucas/PS/PB13B/ps2852.jpg
[22:02:50] <archivist> and the person helping to restore worked on that one in the pic
[22:03:28] <pjm_> i went up to the science museum in london a couple of weeks back, its not anywhere near as good as it used to be
[22:03:32] <pjm_> really dumbed down
[22:03:45] <JymmmEMC> looks like a teletype machine
[22:04:43] <archivist> for IO yes JymmmEMC
[22:05:00] <JymmmEMC> ah
[22:05:13] <archivist> cabinets at window is the computer
[22:05:55] <JymmmEMC> ah
[22:06:17] <archivist> late 1950's
[22:06:18] <JymmmEMC> "That's dumb! Why didn't they use 32" LCD widescreen instead?"
[22:09:33] <archivist> prabably have one of that particular computers books "803 computer library of programmes volume 2"
[23:10:56] <alex_joni> good night all