#emc-devel | Logs for 2008-01-03

[01:35:16] <jepler> cradek: hm, the gcode tests have bitrotted .. they all fail because the output is trivially different than the last time the expected files were updated
[01:35:25] <jepler> N..... SET_ORIGIN_OFFSETS(0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000)
[01:35:25] <jepler> + N..... SET_FEED_MODE(0)
[01:35:25] <jepler> + N..... SET_FEED_RATE(0.0000)
[01:35:25] <jepler> N..... STOP_SPINDLE_TURNING()
[01:35:27] <jepler> + N..... SET_SPINDLE_MODE(0.0000)
[01:35:30] <jepler> N..... PROGRAM_END()
[01:36:11] <cradek> that's from when I fixed M2 for CSS etc
[01:36:59] <cradek> this would not have been my first guess from "realtime checks failing"
[01:37:37] <cradek> (those new results look right to me)
[01:39:31] <cradek> I'll fix those if you'll fix this SCALE documentation (I'm scared by the math mode in lyx)
[01:39:52] <jepler> remind me later
[01:39:56] <jepler> I'm on my dinky laptop and it has no lyx
[01:39:57] <cradek> ok
[01:40:02] <jepler> inifile SCALE stuff?
[01:40:05] <cradek> yes
[01:40:14] <jepler> did I screw it up trying to simplify it?
[01:40:31] <cradek> the example is an inch leadscrew and mm config, and the math doesn't match the answer
[01:40:38] <jepler> oops
[01:40:54] <cradek> a pure inch or mm example is surely better?
[01:41:12] <cradek> oh, in trunk it's right
[01:41:19] <cradek> well different anyway
[01:41:31] <jepler> I inherited the mixed inch/mm example
[01:41:32] <cradek> yeah it's right but different
[01:41:35] <jepler> perhaps I should have had the guts to change it
[01:43:36] <jepler> so you're saying that the number at the end of the math doesn't come out if you work the problem -- not that it's not the same number shown in the sample-ish line, 40000?
[01:44:23] <jepler> looks like to me the examples are the same in 2.2 and 2.3
[01:44:26] <cradek> yes I'm saying the former
[01:44:40] <jepler> three are two -- one for INPUT_SCALE on servos and one for SCALE on steppers
[01:45:07] <jepler> are they both wrong, or just one of 'em?
[01:45:20] <cradek> 2count/1.8 degree * 10rev/inch * 1inch/25.4mm = 157 counts/mm
[01:45:40] <cradek> this one is wrong
[01:45:54] <cradek> degree and rev don't cancel
[01:46:06] <jepler> ah hm
[01:46:24] <cradek> the answer is right, the equation is missing a 360deg/rev
[01:46:25] <jepler> it's also confusing because it's actually finding a reciprocal -- you end up with distance in the denominator
[01:46:48] <cradek> but that's what you want: counts/mm
[01:47:07] <jepler> oh right
[01:47:30] <jepler> units knows rev = turn = circle = 2 pi radian = 6.28...
[01:47:43] <jepler> but maybe my audience doesn't :-P
[01:47:51] <jepler> nor what 'or' means
[01:52:39] <jepler> cradek: so you think I'd be better off with an example that stays within the same units system?
[01:52:55] <cradek> I think so
[01:53:02] <cradek> that'll be 90%+ of the configurations
[02:23:31] <DanielFalck> cradek: Have you found any method of going between autolisp and python? I'm getting into Intellicad/Autocad running under wine. I have been using 'Realize' - Thanks!
[02:37:16] <CIA-20> EMC: 03jepler 07TRUNK * 10emc2/src/emc/usr_intf/stepconf/stepconf.py: two items reported by Harlock: axis test would fail if probe_parport was needed, or if the axis speed was very low (e.g., 200Hz)
[02:37:51] <CIA-20> EMC: 03jepler 07v2_2_branch * 10emc2/debian/changelog: from TRUNK: stepconf bugfixes
[02:37:51] <CIA-20> EMC: 03jepler 07v2_2_branch * 10emc2/src/emc/usr_intf/stepconf/stepconf.py: from TRUNK: stepconf bugfixes
[03:35:55] <cradek> DanielFalck: what do you mean going between?
[03:36:00] <cradek> and you're welcome, glad you like it
[03:36:33] <cradek> (I have a version with a bunch of additional features for the BOSS8 if you care about that)
[03:41:15] <DanielFalck> cradek: hi
[03:41:49] <DanielFalck> I see that there are ways of communicating with other programs using DDE and autolisp (pretty ugly)
[03:42:19] <cradek> sorry I don't know what DDE is
[03:42:41] <DanielFalck> an old scheme like COM or OLE in Windows
[03:42:47] <DanielFalck> but, I don't use windows
[03:42:51] <cradek> I don't either
[03:43:04] <DanielFalck> I was just curious how far you've taken autolisp
[03:43:21] <DanielFalck> I've been playing with it a lot this last week
[03:43:33] <cradek> ideally Realize would be rewritten by an interested person, maybe in a different language (who cares what language), to read DXF and run without Autocad
[03:43:37] <DanielFalck> and there are a ton of autolisp scripts out there
[03:43:53] <cradek> yes they've been accumulating for about 25 years :-)
[03:44:12] <cradek> the fun part is most of them don't work with your version of autocad (whatever it may be)
[03:44:25] <cradek> what version of autocad are you using iwth Realize?
[03:44:33] <DanielFalck> I've had pretty good luck so far
[03:44:37] <DanielFalck> Acad2000
[03:44:46] <cradek> ah mine is much older: R12
[03:44:50] <DanielFalck> I use convertpoly
[03:44:56] <DanielFalck> to make it work
[03:45:00] <cradek> right, glad that still works
[03:45:04] <DanielFalck> R12 was a damn good release
[03:45:12] <DanielFalck> I used it for work for a long time
[03:45:20] <cradek> it's the newest one I've ever used!
[03:45:35] <cradek> learned on R11, R12 added grips
[03:45:44] <DanielFalck> Intellicad and all it's variants is about like acad2000
[03:46:14] <DanielFalck> Intellicad, Progesoft, Bricscad, etc...
[03:46:17] <DanielFalck> ZWcad
[03:46:24] <DanielFalck> and they are all cheap
[03:46:25] <cradek> mumblecad!
[03:46:41] <fenn> grumblecad
[03:46:43] <DanielFalck> Bricscad sells a linux version
[03:47:04] <cradek> I wish there was something FOSS and decent
[03:47:16] <DanielFalck> I have Autocad, Progesoft, and Intellicad working here under Ubuntu and Wine
[03:47:21] <cradek> looks like qcad is the best we'll have for a while I guess
[03:47:35] <fenn> someone just needs to fix brlcad so it's actually a cad program
[03:47:36] <cradek> I'm using R12 on FreeDOS on qemu
[03:47:48] <DanielFalck> In the meantime.... autolisp and intellicad might be fun
[03:48:08] <DanielFalck> I have qemu and win98 in a little box on this machine
[03:48:21] <cradek> R12 is new enough to have output for plotting, a vesa video driver, and a microsoft mouse driver. That means it will work forever which is very nice.
[03:48:35] <cradek> have -postscript- output
[03:49:16] <cradek> qemu is pretty slow to run windows isn't it?
[03:49:26] <DanielFalck> mine works pretty well here
[03:49:37] <DanielFalck> this is a 3 Ghz intel box
[03:49:46] <cradek> that's gotta help
[03:49:52] <DanielFalck> autocad under wine works great
[03:49:59] <DanielFalck> it's very snappy
[03:50:16] <DanielFalck> I made autolisp open up some linux applications
[03:50:48] <DanielFalck> back in a bit.
[03:50:50] <cradek> I don't remember what R12/windows3.1 looks like - it might be pretty bad
[04:03:31] <DanielFalck> cradek: post your Boss8 version on your website
[04:03:55] <cradek> I did (but not quite the latest)
[04:05:14] <DanielFalck> ok, I see it now
[04:05:21] <DanielFalck> thanks
[04:05:23] <cradek> the BOSS8 will actually blend adjacent moves in special circumstances - I took advantage of that in my latest hack
[04:05:30] <cradek> do you have one of those machines too?
[04:05:46] <cradek> by default it stops for about 1/4 second at the end of every segment
[04:05:47] <DanielFalck> no, but it's always fun to see how it's done
[04:07:12] <cradek> on the BOSS8, F8 gives you 0 ipm, F8. gives you 8, F8.0 also gives you 8, and F8.00 gives you 80
[04:08:03] <DanielFalck> I used to run a machine with a 'shadow' brand control on it- it was a clone of the Bridgeport Bandit.
[04:08:40] <cradek> bandit is even older than boss isn't it?
[04:08:44] <DanielFalck> yes
[04:08:50] <DanielFalck> it was very primitive
[04:09:09] <DanielFalck> steppers and little lcd windows to watch the g code scroll by
[04:09:19] <DanielFalck> this machine at least had servos
[04:09:35] <cradek> this has a 2x20 lcd as the only user interface too... it's pretty bad
[04:09:37] <DanielFalck> but it was funny how the company made it just like a bandit
[04:09:59] <DanielFalck> that sounds very familiar...
[04:10:01] <cradek> an unlicensed clone?
[04:10:18] <DanielFalck> I don't know if it was unlicensed.
[04:10:20] <cradek> can't show X,Y,Z at the same time, you have to pick which you want!
[04:10:31] <DanielFalck> The company is 'Shadow' out of Montana
[04:10:53] <DanielFalck> I sure wouldn't want to go back
[04:12:31] <DanielFalck> so your GPL license statement in brealize.lsp must have been copied out of TTT
[04:12:32] <cradek> I can't even find a picture of a bandit - they must have mostly all died before the internet caught on
[04:12:43] <cradek> oops is it wrong?
[04:12:58] <DanielFalck> it refers to TTT instead of brealize
[04:13:08] <cradek> haha oops
[04:13:56] <DanielFalck> I saw a Bridgeport with a Bandit for sale 2 years ago, when I was looking for my mill. It didn't work (Y axis was toast) and the guy still wanted $4500
[04:14:35] <DanielFalck> I decided to pass and I bought mine with servos, 4th axis, tooling, and a Centroid control for that same price
[04:14:39] <cradek> fixed, thanks
[04:14:43] <DanielFalck> cool
[04:15:04] <cradek> wow, $4500 is very ... optimistic
[04:15:21] <DanielFalck> back to the Intellicad/Autocad lisp thing- I think it's a viable option for hobbiest
[04:15:43] <DanielFalck> you can download Progesoft and use it for free
[04:15:57] <DanielFalck> they just want to charge for the Pro version
[04:16:16] <DanielFalck> Intellicad2000 , which was a freebie, is still floating around the web
[04:17:30] <DanielFalck> I like the idea of FOSS for CAM too, but it's good to have options now for people
[04:17:37] <cradek> true
[04:18:33] <DanielFalck> I still see people talking about dual booting and using windows, and it kind of bothers me
[04:18:49] <DanielFalck> so, get everyone used to Linux a bit at a time...
[04:31:30] <cradek> DanielFalck: do you know anything about the wireless (IR) renishaw probes?
[04:31:39] <DanielFalck> no
[04:31:58] <DanielFalck> I've heard that they get sold on ebay - after they are crashed though
[04:32:36] <cradek> I bet so
[04:32:49] <cradek> I'm looking for the receiver part, have the probe
[04:32:55] <cradek> (or, information about how it works)
[04:33:28] <DanielFalck> sorry, I've never run across that.
[04:33:35] <cradek> np, thanks
[04:34:08] <jmkasunich> it's not reassuring to get the "unexpected realtime delay" message 2 minutes into a 60 minute job
[04:34:26] <cradek> jmkasunich: ouch, you're going to have to fix that...
[04:34:41] <cradek> which one did you get?
[04:34:47] <jmkasunich> task 1 IIRC
[04:35:06] <jmkasunich> it only prints once per run of EMC, right?
[04:35:13] <jmkasunich> so I really don't know how often its happening
[04:35:18] <cradek> I think that's right
[04:35:59] <jmkasunich> well, I'm gonna run this part tonight with fingers crossed, and worry about it later
[04:36:13] <jmkasunich> my stepson is leaving tomorrow, it either works or it doesn't
[04:36:31] <cradek> at very slow feed you might survive a glitch
[04:36:36] <cradek> turn down FO so rapids are slow too
[04:36:45] <jmkasunich> I dropped accels down quite a bit
[04:37:04] <jmkasunich> if I turn down FO, it will take all night
[04:37:59] <jmkasunich> the lettering has very few rapids (one per letter mostly)
[04:38:18] <jmkasunich> I think I'll take my chances there, and slow it down for the starts (more rapids there)
[05:00:30] <DanielFalck> cradek: in autolisp- you can load other applications like this: (startapp "/usr/bin/gedit")
[05:00:48] <DanielFalck> or : (startapp "/usr/local/bin/vapt" "/home/dan/apt360/examples/test.apt")
[05:01:18] <DanielFalck> So, I got this crazy idea that I can interact with other programs (python of course) with autolisp
[05:01:29] <DanielFalck> in a crude way- using files
[05:02:14] <DanielFalck> sorry for the non-emc content, but I will open EMC with autocad too : )
[05:06:26] <fenn> you could write out some xml
[05:08:43] <DanielFalck> sorry for my ingnorance, what are you using xml for ?
[05:09:07] <fenn> as an easy way to dump your autolisp data structures
[05:09:24] <fenn> then they can be parsed with a standard xml parser
[05:09:35] <fenn> (in the target python program)
[05:09:57] <cradek> DanielFalck: since I use dos autocad there is no such thing as another program
[05:10:31] <cradek> the easiest way to dump a lisp data structure is to write out the lisp :-)
[05:10:42] <DanielFalck> hmm. ok.
[05:12:30] <jmkasunich> cradek: in axis I can hit the number keys to set FO, right?
[05:12:42] <cradek> yes while running that is what the numbers do
[05:12:59] <jmkasunich> 0 = 100, 1 = 10, 2 = 20 ... 9 = 90 ?
[05:13:06] <cradek> yes
[05:13:09] <jmkasunich> thanks
[05:13:20] <cradek> there's a nice key reference in the help menu
[05:18:13] <jmkasunich> right now I'm trying not to touch it
[05:18:43] <jmkasunich> the touchpad mouse interprets a quick double tap as a click, I don't want to accidently stop the program or something
[05:45:43] <jmkasunich_> jmkasunich_ is now known as jmkasunich
[06:44:12] <jmkasunich> cradek: would you be surprized if axis + xorg were using 100% of the CPU for a long time at "file open"?
[06:52:28] <jmkasunich> never mind, I found the problem
[06:52:45] <jmkasunich> used the same O number for two loops - result = infinite loop
[13:48:25] <CIA-20> EMC: 03jepler 07TRUNK * 10emc2/docs/src/config/ini_config.lyx: fix up SCALE examples based on irc discussion
[13:49:37] <CIA-20> EMC: 03jepler 07v2_2_branch * 10emc2/docs/src/config/ini_config.lyx: from TRUNK: fix up SCALE examples
[14:36:37] <micges> my problem is:
[14:37:10] <micges> I have 3 cnc machines with EMC+AXIS
[14:37:24] <micges> two mills and laser
[14:37:59] <SWPadnos> that doesn't sound like a problem :)
[14:38:11] <micges> and problem is that my machines doesnt have repetivity
[14:39:09] <SWPadnos> I'm not sure I understand what you're saying
[14:39:34] <micges> for egzample few days ago I had send message that EMC 2.2.2 have joint following errors on starting from line
[14:40:15] <SWPadnos> ok - I remember seeing that message
[14:41:24] <micges> I set up program, run, and stop, select some vector and run, then joint errors is showing
[14:42:00] <micges> this was on oryginal AXIS + EMC 2.2.2
[14:42:05] <micges> on 2.1.7
[14:42:13] <micges> it wasnt
[14:42:36] <micges> but after resetup install of EMC
[14:42:42] <micges> first 2.1.7
[14:43:18] <micges> and after one day I installed again 2.2.2 and restart PC
[14:43:33] <micges> the problem dissapeared :/
[14:44:08] <SWPadnos> I don't think I can help you. stay on IRC and hopefully someone who can will show up soon
[14:44:09] <micges> I'm totally have no idea where to start searching what was wrong :(
[14:45:03] <micges> ok thanks
[14:45:13] <SWPadnos> what hardware are you using? (steppers or servos, which servo card if any ...)
[14:45:30] <micges> wait
[14:46:26] <micges> servodrivers BALDOR and hardware step generators connected to parallel port
[14:47:17] <SWPadnos> which parallel port step generator?
[14:47:27] <SWPadnos> or do you mean software step generation?
[14:47:57] <micges> its silly becouse on 2.1.7 everything is ok idependent what I'm doing with EMC installation
[14:48:43] <micges> step generators based on AVR atmega128
[14:49:08] <micges> own production based on PLUTO
[14:49:18] <SWPadnos> ok - interesting
[14:49:56] <SWPadnos> you may need to use halscope to look at the position feedback signal when you try to restart
[14:50:25] <SWPadnos> also position command
[14:52:06] <micges> ok
[15:11:25] <jepler> I'd like to add mathematics to halcmd, e.g., setp stepgen.0.maxvel (1.05 * [AXIS_0]MAXVEL)
[15:11:50] <jepler> I can't see any existing uses of () outside of comments in our sample hal files
[15:11:59] <jepler> anybody anticipate a problem with that notation, or want to suggest an alternative?
[15:13:17] <cradek> [insert troll]
[15:14:41] <jepler> you don't think it's a good idea?
[15:15:14] <cradek> I dunno
[15:15:37] <cradek> inventing a new scripting language is usually not the right answer
[15:15:53] <cradek> I don't know what the right answer is if you already have 75% of the new scripting language you want already invented
[15:16:01] <cradek> already already
[15:17:10] <jepler> yeah
[15:18:00] <jepler> I've also exposed halcmd to tcl, but then you don't have the halcmd-specific inifile notation
[15:18:51] <cradek> yeah that's a nice feature we have
[15:19:03] <jepler> you'd have to write something else .. $AXIS_0(MAXVEL) or $ini(AXIS_0,MAXVEL) or [ini AXIS_0 MAXVEL] are some of the simplest spellings I can imagine, but which are legal tcl
[16:03:11] <SWPadnos> I thought () were used for environment vars
[16:03:26] <SWPadnos> or is that curly braces?
[16:04:06] <cradek> % tclsh
[16:04:06] <cradek> % puts stdout $env(LANG)
[16:04:06] <cradek> en_US.UTF-8
[16:04:14] <SWPadnos> I mean in halcmd
[16:04:31] <cradek> oh I didn't know about that feature
[16:06:28] <SWPadnos> hmmm. ok - parens are significant, but not that way
[16:06:43] <SWPadnos> you can use parens to do the eqivalent of token pasting in C
[16:07:30] <SWPadnos> they're like quotes - so setp parport.$(PARPORTNUM).pin-12-in-invert 1
[16:08:17] <SWPadnos> will work correctly - the envvar is PARPORTNUM (otherwise it would be PARPORTNUM.pin-12-in-invert)
[16:08:27] <cradek> interesting
[16:09:42] <SWPadnos> I guess I could have used quotes there instead, huh?
[16:45:15] <jepler> $VAR accesses a tcl var. $ARR(IDX) accesses an entry in a tcl associative array (strings need not be contiguous integers). There are a few other bits of syntax for variables in tcl, but those are the basics
[16:45:29] <jepler> env is a built in array which holds the environment
[17:29:31] <jepler> I could have sworn I sketched out what it would take to let task dlopen() a behavior library, and waved my hands and said we could provide one that made all the customizable stuff into python calls..
[17:29:35] <jepler> can't find it now
[17:51:18] <alex_joni> jepler: was it task? or canon?
[19:48:21] <alex_joni> hmmm.. since when has SF support for wiki's?
[22:50:46] <maddash> bug: homed flags NEED to be cleared after an estop
[23:50:49] <jmkasunich> says who?
[23:51:51] <SWPadnos> heh