#emc-devel | Logs for 2009-06-11

[01:52:42] <cradek> jepler: what does it do?
[01:56:32] <SWPadnos> I imagine it more or less ignores the g1 line
[01:57:06] <cradek> yeah...
[01:58:45] <SWPadnos> that kind of makes sense though, the "null" X word is equivalent to a line with F only, which is explicitly ignored
[02:05:16] <cradek> it's obviously a corner case, but I don't think I agree -- it is a move that is supposed to take one minute. it is an extremely short move, however.
[02:05:46] <cradek> also, it's almost surely a programming mistake to write it
[02:06:22] <cradek> I'm loathe to define "too short" so a "move is too short in G93 mode" error is right out
[02:07:07] <SWPadnos> heh
[02:07:23] <SWPadnos> EDM guys would probably have a different definition than the rest of us
[02:07:30] <cradek> yep
[02:08:52] <SWPadnos> well, let's see if I bork these embedded machines totally
[02:09:19] <SWPadnos> upgrading all upgradable packages on a dapper machine
[02:09:28] <SWPadnos> at least we still have bfload in EMC2 2.3
[02:49:35] <SWPadnos> hmmm. so what exactly does the file "/etc/emc2/Makefile.modinc" do for me?
[02:50:01] <SWPadnos> the upgrade tells me that I (or a script) modified it
[02:50:17] <SWPadnos> it must have been when I did a make install
[08:27:57] <CIA-48> EMC: 03micges 07TRUNK * 10emc2/src/emc/rs274ngc/rs274ngc_pre.cc:
[08:27:57] <CIA-48> EMC: make sure that canon path control mode is equal motion path control mode
[08:27:57] <CIA-48> EMC: after errors
[09:00:04] <christel> [Global Notice] Hi all, In a couple of minutes we will be performing a clear-out of the Services database, removing deadwood from the NickServ and ChanServ databases. This will slow down services and could temporarily flood them off. It shouldn't take particularly long though! Thank you for your patience and for using freenode!
[09:11:47] <christel> [Global Notice] Hi all, The services prune is complete! If your project is noticing issues with channels in your namespace, please get in touch with freenode staff and we'll sort you out swiftly! Thank you for using freenode and have a great day!
[12:23:21] <jepler> cradek: that line of gcode seems to complete immediately
[12:31:44] <jepler> micges: your last change made the interpreter tests fail.
[12:32:17] <jepler> micges: when you change the output of the interpreter, you have to verify that the changes to the interpreter tests are the expected changes and if so check in new 'expected' results
[12:35:43] <jepler> micges: '. scripts/emc-environment', then 'runtests' in the top emc2 dir. Look at failed tests. For each one, 'cd' to it, then 'diff -u expected result'. If the change in output is expected in view of your change, then 'cp result expected; cvs commit expected'
[12:36:22] <jepler> I am a bit surprised by the change in output, because it shows a nonzero tolerance value. I thought the power-on tolerance value was the same as G64 without P specified.
[12:39:27] <BigJohnT> jepler: is cvs down at the moment?
[12:39:39] <jepler> BigJohnT: it should be fine
[12:39:50] <BigJohnT> let me try again
[12:41:37] <jepler> BigJohnT: there's something odd going on; I'm also having trouble connecting to the cvs server from outside my home network at the moment.
[12:41:47] <BigJohnT> ok, thanks
[12:42:52] <jepler> thanks for the alert
[12:46:05] <skunkworks_> jepler: what pullys did you end up getting?
[12:50:38] <jepler> BigJohnT: check again please
[12:51:41] <jepler> skunkworks_: MXL 1/4" with 12 and 28 teeth (2.33:1 reduction)
[12:52:46] <jepler> skunkworks_: It should give a SCALE of 74666.667, rapids around 2in/sec, and 20 oz-in continuous torque at that speed.
[12:53:30] <skunkworks_> nice!
[12:55:11] <skunkworks_> that is a tenth of a tenth :)
[12:56:14] <skunkworks_> it will be interesting to see what kind of accelleration you will get. Your going to be running 3A?
[12:56:51] <skunkworks_> The spec calculates out to a little over 2A continuous - doesn't it?
[12:59:15] <BigJohnT> jepler: cvs is now working :)
[13:00:17] <CIA-48> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/common/Integrator_Concepts.lyx: add info on rtai and acpi
[13:00:58] <CIA-48> EMC: 03bigjohnt 07v2_3_branch * 10emc2/docs/src/common/Integrator_Concepts.lyx: add info on rtai and acpi
[13:01:20] <BigJohnT> * BigJohnT wanders off to work
[13:09:27] <jepler> BigJohnT: thanks again for the alert; I'm not sure what was wrong, but bouncing my dsl connection seems to have fixed it
[13:27:08] <SWPadnos> jepler, abs has a sign output, does that alter the HAL setup you were thinking of? (for the list user who's asking about mesa vs. pluto)
[13:40:25] <jepler> SWPadnos: yes, it probably makes it a bit simpler -- no need for the comparator
[13:40:28] <jepler> (since when?)
[13:42:27] <SWPadnos> it's been at least 2 years, I guess
[13:43:03] <SWPadnos> hmmm. yeah, looks like it's "since the beginning"
[14:08:01] <jepler> SWPadnos: I'll have to remember that
[15:40:49] <skunkworks_> seb_kuzminsky: !
[15:41:01] <seb_kuzminsky> hoi!
[15:44:21] <skunkworks_> seb_kuzminsky: where did you end up finding this pdf? http://highlab.com/~seb/bzr/fabrication/doc/vendor-docs/pittman/lcm_bulletin.pdf
[15:45:30] <seb_kuzminsky> it was on the penn site i think
[15:46:28] <seb_kuzminsky> it was a while ago i dug it up and i dont remember exactly :-/
[15:46:59] <skunkworks_> heh
[15:49:22] <jepler> seb_kuzminsky: good morning
[15:49:36] <jepler> seb_kuzminsky: have you had a chance to get the new firmwares from pcw yet?
[15:49:39] <jepler> he reported that mail to you was bouncing
[15:49:55] <seb_kuzminsky> he's mailed it to me but i havent checked it out yet
[15:50:09] <seb_kuzminsky> this is the first i've heard that my incoming mail is bouncing :-(
[15:51:04] <seb_kuzminsky> i've been chasing the bugs that cmorley and micges reported
[15:52:34] <seb_kuzminsky> i fixed the encoder-reset bug cmorley reported
[15:52:42] <jepler> emc-devel-2009-06-06.log:08:41 <pcw> Any one know Sebastians Email? Tried to send him a set of updated bitfiles
[15:52:45] <jepler> emc-devel-2009-06-06.log:08:41 <pcw> with the HM2 stepgen bug fixed, but its stuck (retrying on name server lookup failure)
[15:54:41] <alex_joni> sounds like highlab went awoll
[15:55:34] <seb_kuzminsky> it's the stupid qwest modem :-(
[15:56:01] <seb_kuzminsky> it crashes, i power cycle it, it dhcp's and qwest changes my IP addr, then i have to wait for the dyndns updater to run
[15:56:08] <seb_kuzminsky> i should turn up its frequency...
[15:56:47] <seb_kuzminsky> i've been trying to reproduce the "hm2 encoder.velocity is NaN" bug that micges reported but i cant get it to do it here
[15:57:25] <cradek> did he say how he did it? I didn't see much information.
[15:57:49] <seb_kuzminsky> all he said was he changed direction, from 20 to -20 and from 200 to -200.
[15:57:54] <seb_kuzminsky> and he said his encoder scale was 1000
[15:58:12] <seb_kuzminsky> that's all i know, and i've tested all of that here and cant get it to break
[15:58:24] <cradek> hmm.
[15:58:46] <jepler> he is a nan-magnet
[15:58:50] <seb_kuzminsky> heh
[15:58:58] <seb_kuzminsky> also: he broke the build
[15:59:00] <jepler> he gets them often enough in position feedback that he wants to complicate position-file handling just to "fix" it
[15:59:19] <jepler> I could fix it, but I'm stubborn
[15:59:23] <alex_joni> to "prevent" it
[15:59:23] <jepler> the build, that is
[15:59:26] <alex_joni> jepler: same here
[15:59:42] <seb_kuzminsky> it's good to let new developers that break the build figure out how to fix it
[15:59:43] <alex_joni> I've been waiting for micges to turn up so he shall be enlightened
[15:59:44] <Lerman> cradek: I'm having a CVS update problem. Get "could not open lock file...)
[15:59:49] <seb_kuzminsky> it builds character :-)
[15:59:54] <cradek> Lerman: lemme try
[15:59:56] <Lerman> Or alex: or jepler:
[16:00:03] <alex_joni> seb_kuzminsky: it's only a matter of rebuilding the runtest results
[16:00:10] <seb_kuzminsky> sure
[16:00:22] <jepler> when he reads the logs he'll even find that I gave a step-by-step
[16:00:40] <cradek> hmm...
[16:00:43] <cradek> % cvs up
[16:00:43] <cradek> M configs/5axis/5axis.ini
[16:00:43] <cradek> P docs/src/common/Integrator_Concepts.lyx
[16:00:43] <cradek> P src/emc/rs274ngc/rs274ngc_pre.cc
[16:00:43] <cradek> %
[16:01:01] <cradek> looks ok to me - can you give more details? pastebin what you see maybe?
[16:02:22] <alex_joni> that's usually a serverside problem ..
[16:02:29] <alex_joni> but that should be fine now
[16:02:34] <alex_joni> err.. in this case
[16:02:37] <Lerman> could not open lock file `/cvs/emc2/src/emc/rs274ngc/,interp_reead.cc,':Permission denied.
[16:02:59] <alex_joni> was that a copy/paste?
[16:03:03] <cradek> doing what operation?
[16:03:04] <Lerman> (that would be interp_read.cc)
[16:03:31] <Lerman> cvs commit -m"some message"
[16:03:48] <cradek> oh, you said cvs update problem...
[16:03:56] <cradek> are you sure you're not in an anon checkout?
[16:04:05] <jepler> can you make sure it's not an anonymous checkout? cat CVS/Root in any cvs directory
[16:04:35] <Lerman> Yup. It's anon. Boo hiss.
[16:04:41] <Lerman> Sorry.
[16:04:51] <alex_joni> happens ;)
[16:04:54] <cradek> no problem, everyone has done that :-)
[16:05:41] <jepler> do you need help getting a non-anon checkout or transferring your changes to it?
[16:05:44] <Lerman> I sure appreciate the clarity of that message. CVS sucks... let's get rid of it :-)
[16:06:26] <Lerman> I think I can do that. Thanks.
[16:06:37] <jepler> hah!
[16:07:08] <jepler> I bet git provides equally bad errors
[16:07:18] <cradek> oh surely worse :-)
[16:07:35] <jepler> fatal: git-push is not available for http/https repository when not compiled with USE_CURL_MULTI
[16:08:48] <SWPadnos> well, obviously git thinks you should be doing curls instead of push-ups
[16:09:02] <Lerman> jepler: Thanks for simplifying Ed Nisley's bug report. I'm set to check in the fix.
[16:09:09] <SWPadnos> or you can only do push-ups after you have curls. or something
[16:09:12] <cradek> slick, that was fast
[16:09:17] <jepler> Lerman: great, thanks!
[16:09:21] <SWPadnos> cooool
[16:09:31] <jepler> if you feel confident of the fix, I'll backport it to 2.3 for the next release as well
[16:09:44] <Lerman> How do I run the regression tests (I forgot)?
[16:09:49] <cradek> runtests
[16:09:53] <jepler> they're broken at the moment though
[16:09:55] <cradek> although, micges has them slightly broken
[16:09:59] <jepler> . scripts/emc-environment; runtests
[16:10:06] <jepler> but all the interpreter tests will report failures
[16:10:32] <Lerman> Well, I guess I can (must) skip that.
[16:10:41] <jepler> it will show the difference; if it's just a single added line then all's well
[16:13:02] <cradek> fwiw, I think the tests show micges's recent fix is wrong
[16:13:57] <jepler> cradek: the 0.1 is hardcoded in the standalone interpreter; I suspect sai is wrong
[16:14:11] <alex_joni> cradek: how so?
[16:14:18] <cradek> jepler: ahh ok
[16:14:24] <jepler> saicanon.cc:double GET_EXTERNAL_MOTION_CONTROL_TOLERANCE() { return 0.1;}
[16:14:39] <jepler> /* MGS - FIXME - These functions should not be stubbed out to return constants.
[16:19:40] <jepler> though it *is* wrong, because it doesn't restore the naivecam tolerance
[16:21:26] <CIA-48> EMC: 03lerman 07TRUNK * 10emc2/src/emc/rs274ngc/interp_read.cc: Fix improper evaluation of parameters in skipped code.
[16:31:05] <jepler> Lerman: are you confident enough of the fix that I should backport it to 2.3?
[16:32:00] <Lerman> I'd sure like a bunch of people to run existing programs first. Since there is a pretty simple work around, I don't see much urgency for a backport.
[16:32:51] <jepler> what's the workaround?
[16:33:08] <cradek> the gcode programs we've put in the regression tests are much more complicated than the usual thing you'll get people to run by just asking
[16:33:44] <Lerman> Just define the parameter by setting it to some value in a context where it will be executed.
[16:39:15] <jepler> aha
[16:39:16] <jepler> bbl
[18:02:16] <cradek> I wonder if Ed's new article will use the new cutter comp.
[18:02:44] <skunkworks_> Lerman was just wondering about cutter comp the other day. (how complete it was)
[18:02:52] <SWPadnos> is it included in 2.3?
[18:02:55] <cradek> yes
[18:03:02] <cradek> Lerman: complete!
[18:03:34] <skunkworks_> cradek: again - nice work - and on coordinate rotation!
[18:04:51] <cradek> skunkworks_: thanks :-)
[18:05:46] <cradek> rotated g18/g19 arcs still don't preview right in AXIS, but I guess I don't feel like rewriting that - I just hope nobody will notice.
[18:05:55] <skunkworks_> heh
[18:06:19] <cradek> the motion is right (!!), it's just the preview that's bogus
[18:06:45] <skunkworks_> I really doubt many people use g18,g19 arcs. (mostly small line segments I bet)
[18:06:57] <skunkworks_> unless they are writing something by hand.
[18:07:07] <cradek> yeah, on mills at least
[18:07:13] <cradek> g18 arcs on lathes are common of course
[18:07:17] <skunkworks_> right
[18:07:32] <cradek> but if someone rotates the XY plane on a lathe they deserve what they get (and no, I don't know what that is)
[18:07:51] <skunkworks_> heh
[18:09:22] <jepler> cradek: with any luck, you can get ellipses that way
[18:09:46] <cradek> * cradek shivers
[18:10:00] <jepler> and if you rotate 60 degrees, diameter mode
[18:10:24] <cradek> haha
[23:02:21] <SWPadnos> so here's an "interesting" thing that happened when I upgraded EMC2 on a dapper system
[23:02:46] <SWPadnos> the original install had been with the dapper liveCD
[23:03:08] <SWPadnos> I then installed the (now old) experimental SMP kernel cradek made (2.6.20-14 I think)
[23:03:45] <SWPadnos> I then compiled and installed emc2 again, since I had my own components that had to be used (RT drivers and userspace comps)
[23:04:35] <SWPadnos> I just upgraded to 2.3.x last night (changed the repos and used synaptic), among many other updates
[23:05:43] <SWPadnos> when I run emc from a "generic" terminal prompt (no emc-environment), I still get emc2 pre-2.2 CVS head
[23:06:19] <SWPadnos> from the menu, it seems that 2.3 is running, since it complains about not having the right RT kernel (still booting into the SMP one)
[23:07:20] <SWPadnos> oh hmmm. and there's no uninstall target in pre-2.2
[23:17:38] <jepler> SWPadnos: there's never been a "make uninstall" target in emc2
[23:17:47] <jepler> either use RIP or use a packaging system
[23:17:52] <SWPadnos> I thought a (bad) one was added at some point
[23:18:00] <jepler> not afaik
[23:18:04] <SWPadnos> oh, ok
[23:18:10] <jepler> if there is, I'll take it out
[23:18:12] <SWPadnos> well there still isn't one in pre-2.2 :)
[23:20:21] <SWPadnos> I can see the problem now of course - /usr/local/bin is before /usr/bin in the path, and 2.3 got installed into /usr/bin
[23:21:02] <SWPadnos> I would have expected removal of the older emc to remove my make install-ed files though, since theoretically they should have gome to the same locations as the package of the same version did