#emc | Logs for 2008-12-05

[05:33:39] <LawrenceG> hello washington
[06:04:20] <Lerman_______> Lerman_______ is now known as Lerman
[06:06:46] <Lerman_______> Lerman_______ is now known as Lerman
[06:09:11] <maddash> I'm so bored
[06:09:34] <seb_kuzminsky> you could martinize your shorts
[06:09:46] <maddash> 'martinize'?
[06:10:01] <seb_kuzminsky> i think it's some kind of patented dry-cleaning method
[06:10:06] <seb_kuzminsky> the quote is from ren & stimpy
[06:11:01] <maddash> meh
[06:11:27] <seb_kuzminsky> you could play with your magic nose goblins
[06:13:05] <maddash> I have this strong urge to program something, but i don't know what
[06:13:21] <seb_kuzminsky> i'd really like an open-source cad program...
[06:13:28] <seb_kuzminsky> solid modeling
[06:13:33] <seb_kuzminsky> with good gd&t support
[06:14:55] <seb_kuzminsky> or how about a "servoconf" program for emc2? might be easier than a cad program
[06:15:07] <maddash> my attention span isn't long enough to do a cad
[06:15:39] <maddash> is that an analogue to stepconf?
[06:15:42] <seb_kuzminsky> ok, servoconf it is ;-)
[06:15:47] <seb_kuzminsky> right
[06:20:29] <maddash> why does this segfault? main() { char *s="hi"; *s="j"; }
[06:20:56] <seb_kuzminsky> should be *s='j';
[06:21:08] <seb_kuzminsky> "j" is a pointer to char, but *s is a char
[06:21:12] <seb_kuzminsky> 'j' is a char
[06:22:08] <maddash> but why does setting *s to the address of "j" segfault?
[06:22:28] <SWPadnos> it might work (with a warning) if you set s="hello" instead of "hi"
[06:22:41] <SWPadnos> because 2 characters is smaller than a pointer
[06:22:56] <SWPadnos> (or 3, with the terminating NULL)
[06:23:48] <maddash> ooohhhh
[06:24:50] <seb_kuzminsky> i dont think that's why
[06:25:09] <SWPadnos> it could also be that "hi" is allocated in a read-only memory segment
[06:25:11] <seb_kuzminsky> "char c = 1234567878;" doesnt segfault, it truncates
[06:25:16] <seb_kuzminsky> SWPadnos: i think that's it
[06:25:16] <maddash> 1 pointer on my 32-bit system is 32bits wide
[06:25:38] <SWPadnos> seb_kuzminsky, it's not that problem
[06:25:41] <seb_kuzminsky> s points into static data, the assignment tries to write to it
[06:25:50] <SWPadnos> it's not necessarily static
[06:26:19] <SWPadnos> it's constant and initialized, but there's no guarantee that it won't be changed (in the code we saw)
[06:26:31] <SWPadnos> err - initialized with a constant
[06:27:51] <seb_kuzminsky> http://www.eskimo.com/~scs/c-faq.com/decl/strlitinit.html
[06:28:12] <maddash> I hope that I didn't start a fight
[06:28:28] <seb_kuzminsky> you didnt
[06:28:39] <SWPadnos> nope. I just tested and found that my theory is wrong anyway :)
[06:29:49] <SWPadnos> yep, by default gcc is putting both strings in a .rodata section
[06:30:24] <seb_kuzminsky> wacky fun
[06:31:35] <seb_kuzminsky> well i guess i'll go give myself a haircut, see y'all later
[06:32:33] <maddash> i hope that the emc2head debian builddeps haven't changed since the last time I build it
[06:32:39] <maddash> (in jan 2008)
[06:32:49] <SWPadnos> darned good question :)
[06:33:21] <SWPadnos> they may have changed if you changed versions of debian though
[06:33:49] <maddash> <-- lenny
[06:34:36] <SWPadnos> ok. only you know if that's the same as you used in January
[06:36:43] <maddash> yes.
[06:36:52] <SWPadnos> yay! it should still work then :)
[06:37:01] <SWPadnos> TRUNK or 2.2.x?
[06:37:31] <maddash> oh crap, I dl'ed the wrong tgz version
[06:37:49] <maddash> SWPadnos: er, head?
[06:37:58] <SWPadnos> heh
[06:38:25] <SWPadnos> HEAD is the head of a particular branch. TRUNK is the development branck, and v2_2_branch is the 2.2.x branch
[06:38:37] <maddash> geez
[06:38:44] <SWPadnos> you probably mean TRUNK (ie, what you get when you ask for something with no tags)
[06:38:55] <maddash> I just selected "HEAD" from the dropdown menu @ http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/?only_with_tag=HEAD
[06:39:24] <SWPadnos> ok, I'm not sure exactly what that does with viewCVS, but you're probably getting TRUNK
[06:39:29] <maddash> so, HEAD's the right one?
[06:39:39] <SWPadnos> no tag would also be correct
[06:40:21] <SWPadnos> lemme check - one sec
[06:40:23] <maddash> "tag"? there's no item named "tag"
[06:40:42] <SWPadnos> um. ... only_with_tag=HEAD ...
[06:41:33] <SWPadnos> ok, yes. HEAD should be fine
[06:41:41] <maddash> * maddash 's firefox requested HEAD from the server...heheh...
[06:42:26] <SWPadnos> this is what I get: http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/emc2.tar.gz?tarball=1
[06:42:44] <SWPadnos> the tarball link in the emc2/ dir
[06:42:47] <maddash> that's the default option, right?
[06:43:07] <SWPadnos> yes
[06:43:26] <maddash> I tried submitting fixes for homing using that, and jepler scolded me
[06:43:58] <SWPadnos> I'd have to see what you did wrong before siding with either of you :)
[06:44:03] <SWPadnos> err, I mean what happened
[06:50:08] <SWPadnos> well, good night. have fun with it
[06:52:10] <maddash> gnite
[12:19:03] <archivist> one for you lot in the states ebay 180311726072
[12:25:29] <BigJohnT> I don't think my cat would like them :)
[12:26:21] <archivist> but a fat cnc mill would
[12:52:20] <micges> I have:
[12:52:20] <micges> emcTrajSetAxes failing: axes=4 axismask=107
[12:55:32] <alex_joni> micges: I don't have that :)
[12:55:53] <alex_joni> what is your ini settings (number of AXES, and COORDINATES..)
[12:57:44] <micges> emcTrajSetAxes failing: axes=4 axismask=87, and I want XYZV
[12:58:11] <micges> axes=4
[12:58:17] <micges> COORDINATES = X Y Z V
[12:58:28] <micges> 2.2.7
[13:07:08] <alex_joni> axes = 4 is worng
[13:07:11] <alex_joni> wrong even
[13:07:21] <alex_joni> you need 8
[13:08:22] <micges> ?
[13:08:50] <micges> I want 4 not 8, what about rest of not used axes ?
[13:09:03] <alex_joni> x=1,y=2,z=3,a=4,b=5,c=6,u=7,v=8
[13:09:16] <alex_joni> you need a number that includes the last axis you need
[13:09:34] <alex_joni> in your ini you still use only 4 [AXIS_*] sections, and the same in your hal file
[13:10:16] <micges> so if there is no [AXIS_*] section, axis will be not active ?
[13:10:42] <micges> (not enabled)
[13:11:30] <alex_joni> you need AXIS_0 (for X), AXIS_1 (y), AXIS_2 (z) and AXIS_8 (v) in your case (I think)
[13:11:44] <alex_joni> this part of code/configuration is not something I like how it is at the moment
[13:12:25] <micges> me too
[13:12:36] <micges> it works, thanks
[13:36:33] <alex_joni> it is better than it used to be though
[14:12:23] <alex_joni> sounds like myNc is not going anywhere in the future.. http://gcode.blogspot.com/
[14:20:36] <Guest113> why is he writing is own realtime?
[14:20:42] <Guest113> Guest113 is now known as skunkworks_
[14:33:50] <jepler> he seems to be very much a "do it from scratch" person
[14:34:58] <cradek> yeah, wow.
[14:35:08] <jepler> but dang if that isn't a nice 3d preview for wire edm / foam visualization
[14:35:31] <archivist> yup I like that pic
[14:35:58] <cradek> ... and make -mync real-time kernels an option for RTAPI. I've investigated into it recently, and it looks perfectly possible.
[14:36:12] <cradek> this could be cool.
[14:37:02] <archivist> looks like I may have to implement fusee code soon cradek
[14:37:21] <cradek> cool!
[14:37:28] <cradek> on your 4 axis mill it wouldn't be hard at all
[14:37:39] <jepler> ooh looks like cradek just offered to write the code
[14:37:56] <cradek> archivist is perfectly capable - I've seen his youtube videos.
[14:38:03] <archivist> I may get a lathe to turn the blank then finish on the mill
[14:38:18] <cradek> you can stop over and use mine anytime
[14:38:32] <archivist> hehe 3k miles
[14:38:38] <archivist> each way
[14:38:43] <cradek> you could do the whole thing on the lathe like jmk did
[14:38:59] <cradek> you could probably even use his code directly - it uses a table to make the path
[14:39:08] <archivist> nah need proper start stop no silly run ins
[14:39:19] <cradek> is it clock sized or watch sized that you are making?
[14:39:24] <archivist> clock
[14:39:28] <cradek> ah, that's good
[14:40:06] <cradek> it would be neat to set up measurement of spring force for your exact mainspring, and make it according to that
[14:40:07] <archivist> we are quoting for a batch of English dial railway clock movements
[14:40:46] <cradek> oh not repair? making new movements?
[14:40:48] <archivist> yes spring torque measuring will be "fun"
[14:41:04] <archivist> yup maybe 10off
[14:41:09] <cradek> wow
[14:41:27] <archivist> what cnc's are made for:)
[14:41:38] <cradek> sounds like a very fun project
[14:42:03] <archivist> did you see the ebay link I left for you yesterday
[14:42:19] <cradek> no
[14:42:33] <cradek> looking
[14:42:53] <archivist> item 200282170878 up your street
[14:43:04] <archivist> well you make your own
[14:44:06] <cradek> neat
[14:44:28] <cradek> I don't care for numitrons though - they are incandescent filaments so they have a short life
[14:45:33] <cradek> look at the rows of diodes! to multiplex incandescents (which conduct both ways) you need a mess of diodes
[14:45:40] <cradek> one of the benefits of LEDs
[14:46:32] <archivist> I like diodes :)
[14:47:11] <archivist> I should make an electronic clock for me one day
[14:47:32] <cradek> I should make a mechanical clock for me one day
[14:48:40] <alex_joni> archivist: make sure it uses ntp if you do :)
[14:48:54] <cradek> yeah, that is so nice
[14:49:06] <cradek> you guys have an equivalent don't you?
[14:49:07] <archivist> Ive made a couple of k LED clocks for others
[14:49:58] <archivist> alex_joni, or gps and local double oven XTAL
[14:50:30] <archivist> I can get it to a few parts in 10 -9
[15:13:10] <skunkworks_> http://www.cnczone.com/forums/showthread.php?t=69280
[15:14:38] <archivist> ug crappy grounding and poor isolation where needed
[15:16:26] <cradek> wow
[15:16:51] <cradek> absolutely none of the information is there that would let someone actually help him
[15:17:17] <cradek> only the general idea that his setup is not well thought-out
[16:54:00] <piasdom> g'mornin all
[17:20:44] <jepler> why are intel systems so much more popular in the "mini computer" form factor? 37 out of 50 of the "barebone / mini computers" on newgg are intel-based.
[17:24:09] <piasdom> cheaper ?
[17:24:20] <piasdom> abundant ?
[17:30:27] <ewilhelm> lower power consumption and heat dissipation?
[18:12:02] <maddash> fenn: hi, are you there?
[18:12:42] <maddash> http://sourceforge.net/tracker/index.php?func=detail&aid=1901163&group_id=6744&atid=106744 <-- do you use the default configs/sim/axis.ini?
[23:45:16] <JymmmEMC> cradek: what does it output? http://timeguy.com/cradek/01201480300
[23:45:37] <cradek> an irritating answer!
[23:45:45] <archivist> compile it and see
[23:48:01] <archivist> 1.49012e-09 on mine
[23:49:13] <archivist> amazing how many dont understand float errors
[23:49:52] <JymmmEMC> Ok, so what doens't it output" 0.0 ?
[23:49:55] <BigJohnT> if I tie too much chain to my feet I get float errors
[23:50:11] <archivist> which bugs me about float use in EMC, I know it could be out for gear hobbing
[23:51:47] <cradek> EMC2/trunk now uses doubles everywhere
[23:52:26] <cradek> will be in EMC2.3
[23:52:41] <archivist> hehe instead of a sensible page about floats "problems with floats" pulls up loads of whining forum posts
[23:52:55] <anonimasn> archivist: do you really belive that your machine is more accurate then a float?
[23:53:02] <cradek> Results 1 - 10 of about 487,000 for floats suck.
[23:53:16] <anonimasn> I think that rounding errors is insignificant at that point
[23:53:31] <archivist> for hobbing I deed an integral number of rotations
[23:53:34] <cradek> anonimasn: floats get less precise as you move away from 0
[23:53:56] <cradek> on a rotary table that turns the same way for a long time, your position can get pretty choppy
[23:54:01] <cradek> (if position is a float)
[23:54:23] <cradek> now that it's a double, that problem is virtually gone, along with many others
[23:55:09] <anonimasn> cradek: yeah it's alot better.
[23:55:16] <archivist> my cutting spindle will do thousands of revs and the rotary will do hundreds, then I want to pull out and recycle
[23:55:20] <cradek> with floats I think you could have significant problems with only a thousand turns of a rotary axis (360,000 degrees)
[23:56:16] <cradek> on a noninfinite axis like any linear sliding thing, I agree you're unlikely to see problems caused by floats unless your units are very tiny
[23:56:45] <anonimasn> yep that was the point, not there isnt a problem with floats
[23:57:04] <cradek> yeah rotaries are a bigger problem.
[23:57:32] <JymmmEMC> isn't there an equiv of bcmath?
[23:58:32] <archivist> cradek, and I need to rotate z geard with spindle and not bother unwinding eg go straight to 0 deg/start angle
[23:59:04] <cradek> spindle free running?
[23:59:25] <archivist> no
[23:59:51] <archivist> got to be well geared with rotary