#emc | Logs for 2004-11-17

[00:00:02] <rayh> pfred1_: I can't remember her name. Cute though.
[00:00:15] <pfred1_> rayh I always knew you were A #1
[00:00:22] <alex_joni> hey rayh
[00:00:27] <rayh> Yea right.
[00:00:34] <alex_joni> not really auto stuff... more install issues
[00:00:45] <alex_joni> actually preparing things in order for install to work
[00:01:12] <pfred1_> if The Matrix was real everyone would take the "Blue Pill"
[00:01:21] <rayh> alex_joni: This is the jump from run-in-place to a correct install.
[00:01:51] <rayh> I believe that there are a few who would still prefer the red.
[00:02:46] <alex_joni> rayh: we are actually trying to keep run-in-place
[00:02:54] <alex_joni> and get install done
[00:03:15] <rayh> Good. We need run in place for developers and testers.
[00:03:15] <alex_joni> as long as emc2 is not production/stable run-in-place is very usefull
[00:03:23] <pfred1_> unlike this guy I still have a full head of hair! http://money.cnn.com/2004/11/15/news/fortune500/pfizer_ad.reut/viagra_hes_back.03.jpg
[00:03:48] <alex_joni> I know...
[00:04:45] <rayh> I hear the fed made them pull the ad.
[00:04:53] <pfred1_> who has a sunday paper? I need to find a CD sleeve deal ...
[00:07:19] <rayh> All the news that fit to print -- and an ideal wrapping for sandwiches or CDs?
[00:07:48] <pfred1_> rayh you go through the sunday paper every week some chain store running a deal on blank CDs
[00:08:05] <pfred1_> get 50 for anywheres from $0-$5
[00:08:21] <rayh> Oh. I don't do that cause I've got a few thousand bdi blanks.
[00:08:37] <pfred1_> rayh if they're BDIs how can they be blank?
[00:09:03] <rayh> Got them printed with those great images from Rab Gordon.
[00:09:21] <pfred1_> oh you mean they printed to say BDI but nothing is on them?
[00:09:30] <rayh> Right.
[00:09:39] <pfred1_> * pfred1_ uses a magic marker
[00:09:48] <pfred1_> hi tech!
[00:10:29] <pfred1_> so this RTAI stuff they got anything other than this live test suite?
[00:13:25] <les> hi all...just in checking morning email
[00:13:31] <alex_joni> paul_c: how about running all stuff from the configdir?
[00:13:52] <alex_joni> inside emc.run cd-ing to the configdir, and run from there
[00:14:13] <jepler> cradek: Well I think I hacked around the memory leak in PyOpenGL. I'll let 3D_Chips run to completion in live preview before I declare victory.
[00:14:20] <paul_c> and if the ini path is pointing elsewhere ?
[00:15:04] <alex_joni> well then it should make sure it points right
[00:15:14] <alex_joni> relative dirs or absolute dirs
[00:19:29] <alex_joni> I really think defaults inside the ini should not include any dirs
[00:19:40] <alex_joni> and all files should be (get installed) into the same dir
[00:27:35] <cradek> jepler: yay, I guess
[00:32:42] <alex_joni> jepler: working?
[01:22:43] <alex_joni> * alex_joni has gone home...
[02:23:49] <jepler> SIZE/RSS before->after: 37216 / 16M -> 37444 / 16M
[02:24:46] <paul_c> Sounds like you are plotting every point along each segment
[02:27:17] <jepler> actually I try to detect colinearity of successive points, but I don't think it has a big impact on memory usage
[02:27:47] <jepler> I don't know if you read it, but the problem last night was that a bug in PyOpenGL leaked memory horribly, so that the live plot grew to around 1 gig during the plot of chips
[02:27:58] <jepler> so growing only 400k is great
[02:28:44] <paul_c> Was there a fix for the memory leak ?
[02:30:12] <jepler> I traced the leak to the wrapper for glInterleavedArrays and wrote a new one without the bug.
[02:33:18] <paul_c> It looks like I should be able to add pyopenGL to the next BDI - Might need to drop some other packages to avoid growing to a second CD
[02:34:07] <jepler> and what version of Python is on BDI?
[02:34:22] <paul_c> it will be 2.3
[02:35:28] <jepler> let me know and I'll help you through getting the rs274py viewer built for BDI.
[02:36:40] <paul_c> currently have python-2.3.4 installed
[02:38:17] <jepler> will the live plot code that works with emc1 continue to work with emc2?
[02:38:31] <jepler> If there's no "sim" for emc2 I'll have no way to run/test it, since I don't have a mill
[02:39:14] <paul_c> I see no reason why emcplot3d shouldn't work...
[02:40:20] <jepler> so RCS_STAT_CHANNEL and the rest will not change?
[02:40:57] <paul_c> nope
[02:42:46] <paul_c> The rcs interface is a cornerstone of emc
[02:46:13] <jepler> So with emc's rs274ngc parser is there a way to query the current value of parameters for a given line, like I can get the modal codes with rs274ngc_active_g_codes() ?
[02:46:37] <paul_c> yes
[02:47:13] <jepler> so I can find out what is stored in #1001 or another arbitrary parameter number
[02:47:33] <paul_c> Don't think so..
[02:48:44] <paul_c> in emc2, the rs274 interp is wrapped in a C++ class, so much of the internal data is private
[02:48:50] <jepler> the two reasons I wrote my own parser were because I didn't want to have to interface to emc, and because I wanted to be able to query parameter values (so that, for instance, a program can say what RGB color things should be drawn in)
[02:51:29] <paul_c> * paul_c has to disappear for a bit.
[02:55:56] <jepler> looks like in emc1 I could write _interpreter_settings.parameters[101] to get the value of #101
[03:05:42] <cradek> jepler: you don't need a mill to run emc2, but it sounds like you need a realtime kernel.
[03:08:00] <jepler> OK, maybe that's a bar I can meet
[03:08:16] <cradek> File "/usr/bin/gplot", line 617, in main
[03:08:16] <cradek> if k == '-d': emcshow.emlfile = v
[03:08:16] <cradek> NameError: global name 'emcshow' is not defined
[03:08:23] <jepler> oops
[03:09:21] <jepler> should be "emc.emlfile" not "emcshow.emlfile" I guess.
[03:14:03] <cradek> whee, look at the dancing cone
[09:34:01] <dave-e> paul...are you hiding?
[09:34:40] <les> naw I was just talking to him
[09:34:56] <dave-e> he may have seen me coming
[09:35:22] <les> dave...would you say that the Vital card is debugged and fully functional in servo emc?
[09:35:28] <dave-e> am I the only one having trouble compiling the cvs under rc46?
[09:35:53] <dave-e> les...I don't know...that is exactly what I'm working on.
[09:36:08] <paul_c> dave-e: probably.
[09:36:17] <dave-e> hi paul
[09:36:53] <dave-e> I've just tried to compile yesterdays cvs under rc46 using both your procedure and one from ray.
[09:36:59] <dave-e> no joy so far....
[09:37:22] <dave-e> it appears the emcmot is broken....lots of errors
[09:37:51] <paul_c> emcmot hasn't altered for some time...
[09:38:03] <dave-e> OK then something else is wrong.
[09:38:41] <dave-e> I can go back and play so more.
[09:38:59] <dave-e> I tried the procedure you put on the list for Jon..
[09:39:13] <dave-e> is there anything else I should know?
[09:39:19] <paul_c> email me with the process you are using along with some of the errors
[09:40:02] <dave-e> OK ... that will take a while...I did it without saving the log...you know in ultimate trust that it would work.
[09:40:34] <dave-e> 'course my typing isn't perfect either. ... or maybe my understanding of what is supposed to go on.
[09:40:47] <dave-e> will give it another try and save the log.
[09:40:54] <dave-e> les...
[09:41:00] <les> yes?
[09:41:32] <dave-e> On my earlier try at the vital board it had some noise problems...pickup in the flat cables. .. a real headache...
[09:41:47] <les> using diff encoders?
[09:41:49] <dave-e> Abdul made some changes and it should now be OK
[09:42:33] <dave-e> with diff encoders....the problem was emi coming in on the flat cables between the motherboard and the breakout boards.
[09:42:53] <dave-e> so far this version looks much quieter.
[09:42:58] <les> Well...I am trying to get a 4 axis mill to retrofit...need another machine....
[09:43:13] <les> not keen on buying anymore stg boards
[09:43:34] <dave-e> I really cannot say much until I get farther along on the test process.
[09:43:59] <les> I find that the flat ribbon twisted pairs work well
[09:44:11] <dave-e> I'm hoping the vital board set will check out; if so it will be a good choice.
[09:44:20] <les> for sure
[09:44:32] <dave-e> I have some Spectra cable here but have not used it.
[09:44:57] <dave-e> My physical arrangement this time allows for much shorter cables that the earlier setup.
[09:45:06] <les> some ferrite beads here and there might work wonders too
[09:45:40] <dave-e> I think Abdul added some r/c filters to the incoming lines.
[09:45:57] <les> that will work as well
[09:45:59] <dave-e> So far I don't have the noise I had last time
[09:46:03] <les> good
[09:46:32] <dave-e> I'll know a lot more as soon as I get real servo control.
[09:46:39] <dave-e> this time I know what to look for.
[09:46:50] <les> "terminate lines in characteristic impedance" right?
[09:47:20] <dave-e> well I assume Abdul did that kind of thing ... he is a pretty good engineer.
[09:48:01] <dave-e> The boards looked good for homing and tuned well so the noise was really the only problem.
[09:48:42] <dave-e> thanks paul...back to the drawing board...or at least the keyboard. :-)
[09:49:01] <les> oops
[09:49:45] <dave-e> les...I'll touch bases later and bring you up-to-date