#emc-devel | Logs for 2007-10-12

[00:01:16] <skunkworks> hey john. Seems as if you have been busy. :) like you have a life or something.
[00:02:35] <SWPadnos> howdy
[00:02:53] <jmkasunich> yeah, life is busy
[01:18:51] <cradek> hi!
[02:00:24] <SWPadnos> hmmm
[05:56:31] <fenn> swpadnos actually you can turn a polygon without cnc, any even numbered polygon
[05:57:01] <fenn> oh, actually any number. yeah.
[12:36:51] <jepler> did we determine that running without STEPGEN_MAXVEL caused no discernible problems?
[12:38:31] <skunkworks> The little testing I did - the onlything that seemed to need to be set was stepgen_maxaccel
[12:38:52] <skunkworks> (headroom)
[12:41:42] <skunkworks> stepgen_maxaccel set to 0 or a large number cause stalling. vs 20 or so percent higher than max_accel
[12:42:07] <skunkworks> on the hermes
[13:05:05] <skunkworks> The preview window in axis is opengl correct?
[13:28:38] <alex_joni> right
[13:30:36] <skunkworks> THanks alex
[13:30:41] <skunkworks> Its friday
[13:33:40] <jepler> skunkworks: if you're reporting a bug, you should verify that it exists with 'glxgears'
[13:34:16] <cradek> very good idea to not mention any program that isn't on the base installation
[13:36:18] <skunkworks> so I should install gutsy without the rt kernel and just run glx gears?
[13:37:09] <cradek> well, that sounds like a lot of work, but would definitely make sure it's not jepler's fault
[13:38:20] <skunkworks> heh
[13:38:41] <jepler> rebooting with the "vanilla" kernel is also a good idea
[13:39:11] <jepler> for accelerated opengl, a (small) portion of the code is in the kernel as well
[13:39:18] <jepler> should be no need to reinstall though
[13:39:53] <skunkworks> or what about building sim?
[13:40:12] <skunkworks> but glx gears should be a good test.
[13:40:31] <jepler> if it doesn't happen with glxgears then it's not likely to get their attention
[13:40:43] <jepler> right or wrong, they'll blame the app that isn't a part of obento
[13:42:34] <cradek> jepler: you left some SCALE = 1234 0. did you mean to?
[13:47:56] <alex_joni> bbl
[13:51:33] <jepler> cradek: I wasn't thinking about that at the time
[13:54:26] <cradek> thanks
[14:25:43] <fenn> it's actually spelled obentou :\
[14:26:09] <fenn> but its a good distro name i think :)
[14:30:49] <jepler> obento is the transliteration I've always seen. wikipedia has "bento" with a redirection from "obento". I've never heard obentou before, but on the first page of google excerpts I see that "Japanese runch ='OBENTOU'"
[14:31:17] <jepler> hm, maybe bento is the transliteration I've seen, but the 'obento' distribution joke has been floating around here at the office for awhile..
[14:32:44] <skunkworks> was aram at the cncworkshop?
[14:32:50] <jepler> yes
[14:32:59] <jepler> you didn't get the chance to meet him?
[14:33:01] <fenn> he's the annoying guy
[14:33:19] <skunkworks> I think I did - he tried to use my laptop to hook up to his drive.
[14:33:26] <skunkworks> he said my laptop was broken ;)
[14:33:35] <skunkworks> he had a strong accent?
[14:33:41] <fenn> yep
[14:34:03] <skunkworks> trying to get emc to work with his drive? he had a little test setup running.
[14:34:23] <fenn> yeah i think he should just get something easier to interface with
[14:36:59] <skunkworks> fenn: you where not there this year where you?
[14:37:20] <fenn> no
[14:37:23] <cradek> I think he has been trying for over a year to tune some pid
[14:38:10] <skunkworks> I think just in passing - I thought jone figured out the drive must have been set for current instead of velocity or something like that.
[14:38:47] <cradek> I remember there was some windows software involved
[14:39:03] <cradek> in his last message I don't even understand what he wants
[14:39:15] <cradek> he wants to remove most of the gcodes? or something?
[14:39:22] <skunkworks> Yes - the software would not run on my xp machine. He said may portable was broken.
[14:39:43] <cradek> ah
[14:40:09] <jepler> I remember he did something on steves_logging's windows pc that annoyed him (steve) too
[14:40:13] <jepler> installed software or I dunno what
[14:40:20] <skunkworks> heh
[14:41:06] <fenn> cradek: he wants to add extra rotational axes like UVW but for ABC instead of XYZ
[14:41:35] <skunkworks> he is the type you just agree with whatever he said and try to get away.. He had issues with rigid tapping also - something like 'no way it will track well enough not to need floating holders' or something like that.
[14:42:15] <fenn> yes he told me CNC was a waste of time and i should go to medical school
[14:42:35] <cradek> fenn: I think he thinks UVW are incremental?
[14:42:42] <cradek> fenn: haha
[14:42:54] <fenn> well, traditionally UVW is added to xyz position, so with u=0 it will track X
[14:43:07] <fenn> according to tomp
[14:43:13] <cradek> s/traditionally/on some machines/
[14:43:18] <fenn> yeah :\
[14:43:22] <skunkworks> that is what i understand also
[14:43:25] <cradek> /on wire edm/
[14:43:34] <skunkworks> or foam machines.
[14:43:57] <cradek> stuart's making an XYZW machine where W is the table that moves in/out from the spindle
[14:43:58] <fenn> i could care less about uvw really. i think if you want to specify more than one thing moving at a time, you need to have another g-code interpreter running
[14:44:02] <cradek> W is often a knee too
[14:44:09] <skunkworks> I would think that would be an easy filter to add to axis :) if someone really wants it..
[14:44:21] <fenn> cradek: well at least they got the letter right
[14:44:40] <skunkworks> take the reletive uvw and convert to absolute.
[14:45:04] <fenn> its also a quick hack in hal
[14:45:10] <cradek> no it's not
[14:45:12] <fenn> its not?
[14:45:27] <skunkworks> there are issues with accel and vel constraints that hal doesnt see
[14:45:28] <cradek> g1x1u1 / x2u1 / x3u1
[14:45:41] <cradek> notice u doesn't change, but x should be at ... uh I dunno
[14:45:49] <fenn> adding u to x should get what you want
[14:45:52] <cradek> 6?
[14:46:01] <cradek> no it doesn't
[14:46:13] <cradek> 3+1=4, when you should be at 6 I think
[14:46:13] <jepler> x3u1 would presumably be unacceptable
[14:46:21] <cradek> it has to go in the interp
[14:46:33] <fenn> er.. 3+1 = 4
[14:46:54] <fenn> axis.0.* would be at 3
[14:47:12] <cradek> g1x1y1u1 / y2u1 / y2u1 / y2u1
[14:47:15] <fenn> axis.3.* would then be added to axis.0 to get the final value to go to the motors
[14:47:22] <cradek> now x=3 I think
[14:47:23] <jepler> My guess was that this was supposed to draw a square: G1 U1; V1; U-1; V-1
[14:47:35] <cradek> fenn: no you're still not seeing it - it's incremental
[14:47:45] <fenn> its not a constant offset?
[14:48:01] <cradek> for wire it is, for a lathe it's not
[14:48:11] <cradek> sorry I think we're talking about different things
[14:48:25] <fenn> step 1: define the problem
[14:48:26] <cradek> err in my example now x=1+1+1+1=4
[14:48:38] <jepler> I thought we were trying to guess what aram meant
[14:48:54] <jepler> which is not necessarily the same as what any real-world system does :-P
[14:49:00] <cradek> true
[14:49:12] <cradek> there are (at least) two kinds of incremental that have something to do with uvw :-/
[14:50:20] <fenn> in your example the only other way to interpret it is (1+1)+(2+1)+(3+1) = 9
[14:50:56] <cradek> which example?
[14:51:06] <fenn> the first (g1x1u1 / x2u1 / x3u1)
[14:51:35] <fenn> and what happens if you dont call u? is it different than u0?
[14:51:58] <cradek> I have no idea
[14:52:17] <cradek> I just know some lathes do something surprising with u,w and it's something like that
[14:52:29] <cradek> doubt any of them actually agree either
[14:52:59] <cradek> but (and, sorry) my point was that you can't get this kind of monstrosity^Wbehavior with any hack in hal
[14:53:23] <cradek> (my bridgeport actually does something like this too with uvw, fwiw)
[14:54:05] <jepler> in emc, UVW have positions just like XYZABC; kinematics can produce motor positions in a way that makes sense for the machine (for instance, maybe joint 0 = X and joint 2 = X+U makes sense on certain wire machines). "What happens if U isn't specified on each line of gcode" is the same as "What happens if X isn't specified on each line of gcode".
[14:54:59] <jepler> from aram's text, it sounds like he wants "U is like X-but-with-G91-active" which can't be accomplished in kinematics (because "what happens if U isn't specified on each line of gcode" is different)
[14:55:23] <cradek> I agree, and that's a good summary
[14:57:28] <jepler> two corrections: "UVW have axis positions", "kinematics can produce joint positions"
[14:57:44] <jepler> hah
[14:57:44] <jepler> It's a great world we live in when we can communicate even though we don't
[14:57:45] <jepler> agree on how many letters in the alphabet. :-)
[14:57:49] <skunkworks> plus he want incramental abc also
[14:58:13] <jepler> aka 'akA'
[14:58:32] <cradek> akA?
[14:58:44] <jepler> akA incremental rotational motion around X ases.
[14:59:12] <fenn> ak = his initials
[14:59:23] <jepler> oh
[14:59:51] <cradek> ohhh
[14:59:53] <skunkworks> well - that makes sense ;)
[15:00:23] <cradek> what is sintecs?
[15:00:33] <jepler> "syntax"?
[15:00:57] <cradek> oh I bet so
[15:02:26] <cradek> funny. I didn't even catch the 28 that ken joked about
[15:02:56] <jepler> 'Arabic is written from right to left; its alphabet is composed of 28 basic letters'
[15:03:10] <cradek> ah!
[15:06:25] <fenn> do ( and # count as letters? i guess [ would have to be added, but E isnt a g-word
[15:06:46] <cradek> E is used now
[15:06:52] <cradek> (in emc)
[15:06:56] <fenn> by what?
[15:06:59] <cradek> g76
[15:07:15] <fenn> why?
[15:07:22] <cradek> why not?
[15:07:54] <fenn> because then you cant use E for anything else?
[15:08:16] <cradek> why not?
[15:08:30] <fenn> i thought the interp was stupid that way
[15:08:34] <cradek> P,Q are used all over the place
[15:08:41] <fenn> i mean as its own command
[15:08:51] <cradek> I don't understand
[15:08:59] <fenn> like, elliptical blending or something
[15:09:40] <cradek> if only I could post the bunny/pancake
[15:09:42] <fenn> or the Electrical axis
[15:10:17] <cradek> http://www.kurtsimmons.com/images/bunny-pancake.gif
[15:10:32] <cradek> thanks google
[15:11:00] <fenn> http://fenn.dyndns.org/pub/irc/easter_is_cancelled_jpg.jpg
[15:14:54] <fenn> are you sure E is used with g76? i dont see it in the docs
[15:16:15] <cradek> yes, has something to do with tapered entry/exit
[15:16:32] <cradek> the docs you're reading (and maybe all of them) may be out of date
[15:17:06] <cradek> http://cvs.linuxcnc.org/cvs/emc2/nc_files/g76.ngc?rev=1.9
[15:17:27] <fenn> was this done to match some other control dialect or is it arbitrary?
[15:17:40] <SWPadnos> E was the only unused letter
[15:17:53] <cradek> nobody agrees on things like canned cycles
[15:18:01] <SWPadnos> which strangely is true of ls also
[15:18:03] <cradek> I wanted the functionality so I did it
[15:18:46] <fenn> SWPadnos: coincidence?
[15:18:58] <SWPadnos> I thikn not!
[15:19:03] <SWPadnos> err- think
[15:19:06] <SWPadnos> I need coffee
[15:19:30] <fenn> it doesnt use y or z either
[15:19:44] <cradek> fenn: the comments in this file may be the only g76 docs that are up to date
[15:19:51] <cradek> (please fix if you care)
[15:21:25] <fenn> gragrrr spindle start has to be one of these m-codes
[15:22:31] <fenn> s1000 m3 should start the spindle right?
[15:22:42] <fenn> ooo duh
[15:22:49] <fenn> nevermind :)
[15:29:20] <fenn> E is also used by M66
[15:30:39] <jepler> s/care/dare/
[15:30:43] <jepler> M66 came along after G76
[15:31:04] <cradek> I have no idea what M66 is
[15:31:22] <jepler> has to do with external inputs
[15:31:32] <cradek> oh right
[15:31:37] <cradek> I still haven't tried that
[15:31:53] <jepler> I think part of it is unfinished, but I don't remember what :-P
[15:33:23] <cradek> good thing emc is done, because we've all run out of steam
[15:33:58] <cradek> 'can do everything I imagine wanting in the next year or two' = done
[15:34:31] <fenn> go write some good cad software :P
[15:34:46] <fenn> i have some steel beams and i want the ends just so
[15:35:15] <cradek> nope, I have autocad, so I'm not interested in writing cad software
[15:39:48] <skunkworks> we need to get cradek into casting..
[15:40:04] <skunkworks> then maybe there would be a nice 3d cam/cad package..
[15:40:06] <skunkworks> :)
[15:40:14] <cradek> have done some, but I don't really have a need I guess
[15:40:35] <cradek> I'll try lost-foam next time I need something
[15:41:51] <fenn> skunkworks: what's really missing is a parametric modeler. you can make fancy shapes all day long in blender
[15:42:20] <fenn> and anders wallin is making good progress with cam i think
[15:42:33] <skunkworks> I saw some of that.
[15:42:34] <skunkworks> cool
[15:42:47] <fenn> as far as surface contouring goes at least
[15:42:57] <fenn> not counterboring holes or anything
[15:43:15] <skunkworks> I still have not done really any 3d cad.
[16:00:09] <skunkworks> glxgears does it also - on a fresh install
[16:00:34] <cradek> yay (I guess)
[16:00:52] <skunkworks> so - somebody screwed something up ;)
[16:01:11] <cradek> and not jepler
[16:01:16] <skunkworks> right
[16:17:19] <skunkworks> http://www.electronicsam.com/images/KandT/GLXGEARS.png
[16:17:38] <skunkworks> this is just booting off of the livecd - the other computer is taking forever to update.
[16:18:50] <cradek> sure looks wrong to me...
[16:20:39] <skunkworks> it also 'stick' when it is dragged around as you can see.
[16:20:48] <cradek> yep
[16:21:09] <cradek> I bet you should also include the output of "glxinfo" in your report
[16:21:34] <skunkworks> ok
[16:22:55] <skunkworks> really though - it did it on 2 different computers - I think it would be really easy to reproduce.
[16:23:08] <cradek> yeah you would think.
[16:34:23] <skunkworks> http://www.electronicsam.com/images/KandT/GLXGEARS1.png
[18:47:25] <cradek> "The U, V, and W axes produce linear motion in three mutually orthogonal directions. Typically, X and U are parallel, V and W are parallel, and Z and V are parallel." </the docs>
[19:29:18] <alex_joni> 'lo
[19:29:28] <cradek> clear as mud, no?
[19:30:34] <cradek> Q and W are parallel, as are 6 and J, W and V, [ and E, etc. etc.
[19:30:50] <skunkworks> V and W?
[19:30:58] <skunkworks> you mean Y and W?
[19:31:03] <cradek> Z and V?
[19:31:16] <cradek> it's all wrong...
[19:31:20] <alex_joni> Y and V
[19:31:21] <cradek> very funny
[19:31:27] <alex_joni> Z and W
[19:31:35] <cradek> I know what it's supposed to be
[19:31:40] <cradek> I was just sharing the gem I found in the manual
[19:31:42] <skunkworks> hhe
[19:42:46] <alex_joni> shall we guess who put it there?
[20:07:43] <cradek> no that would be a bad idea :-)
[20:09:36] <alex_joni> indeed
[20:27:27] <cradek> (was it me?)
[20:29:50] <alex_joni> nope
[20:30:15] <cradek> oh good
[20:30:33] <alex_joni> * alex_joni ponders about going to bed
[20:32:01] <alex_joni> seem slike a good idea
[20:32:06] <fenn> indeed
[20:32:11] <cradek> goodnight
[20:32:35] <alex_joni> good night all
[20:51:57] <SWPadnos> I wonder if it would be a good thing to add a command to halcmd to turn off error reporting temporarily
[20:52:14] <SWPadnos> so changes like adding/removing the STEPGEN_MAXVEL settings wouldn't be needed
[20:52:44] <SWPadnos> you'd just protect blocks of optional settings with errors on / ... / errors off (or similar)
[20:52:58] <SWPadnos> in the hal file
[22:04:49] <dmwaters> {global notice} Hi all, our main US hub is about to go down for some upgrading. This will cause some major splitting. We will work to get things back together as quickly as possible. Thank you for your patience, and thank you for using freenode!
[23:08:06] <skunkworks> got my green laser. it turns on :) will try to see if it illuminates the sky tonight.
[23:34:40] <LawrenceG> tower this is United 475 heavy... we are being targetted by a bright green dot.... please advise
[23:39:18] <skunkworks> heh
[23:39:54] <skunkworks> there was a csi where someone with a small green laser pointer brought down a passenger jet... And I suppose people believe that.