#emc-devel | Logs for 2009-11-14

[00:44:18] <cradek> it's not hard to beat gitweb IMO
[00:44:41] <cradek> hg has the web part built in, so you can do ad-hoc servers ("hg serve" iirc)
[00:45:39] <jepler> there's 'git instaweb' but gitk is so much better than anything web-based that I don't see why you'd bother
[00:46:14] <cradek> jepler: when sitting next to someone who you want to show something
[00:46:15] <jepler> (and it requires one of a few web servers to be installed, anyway)
[00:46:28] <cradek> but yeah, minor
[20:09:17] <mozmck> what in emc uses python-xml? I ask because that package is going away (gone in ubuntu 9.10)
[20:09:33] <mozmck> the functionality is duplicated in core they say...
[20:12:01] <awallin> pyvcp parses xml files
[20:12:10] <awallin> using minidom ?
[20:13:41] <awallin> is there a way of web-browsing the git repo?
[20:14:26] <micges> http://git.linuxcnc.org/gitweb?p=emc2.git;a=summary
[20:15:31] <awallin> look here for what imports pyvcp uses: http://git.linuxcnc.org/gitweb?p=emc2.git;a=blob_plain;f=lib/python/vcpparse.py;hb=529e0b8712295bc5d77b554c3079064241837072
[20:16:02] <mozmck> so pyvcp would be the only thing that uses it?
[20:18:13] <mozmck> I guess I can check through the other code. I wonder what it would take to change it? ..and if it would be backwards compatible with hardy
[20:18:23] <awallin> I think so
[20:18:43] <awallin> do the "import xml.dom.minidom" etc. lines need to change somehow?
[20:25:04] <mozmck> I don't know for sure, I just found out they remove python-xml from ubuntu 9.10 and up, so by the next LTS we will have to change something!
[20:25:59] <awallin> I am on 9.10 now.
[20:26:16] <awallin> anders@anders-desktop:~$ python
[20:26:18] <awallin> Python 2.6.4 (r264:75706, Nov 2 2009, 14:44:17)
[20:26:20] <awallin> [GCC 4.4.1] on linux2
[20:26:22] <awallin> Type "help", "copyright", "credits" or "license" for more information.
[20:26:24] <awallin> >>> import xml.dom.minidom
[20:26:26] <awallin> >>> import xml.dom.ext
[20:26:29] <awallin> Traceback (most recent call last):
[20:26:30] <awallin> File "<stdin>", line 1, in <module>
[20:26:32] <awallin> ImportError: No module named ext
[20:26:34] <awallin> >>>
[20:26:40] <mozmck> I'm working on a livecd based on 9.10. My main system is 9.10 but I upgraded so I still have python-xml installed.
[20:27:32] <mozmck> no problem importing xml.dom.ext here
[20:28:50] <mozmck> I've got an rtai patched kernel, and rtai and emc .debs for 9.10 on experimental if you're of a mind to try them ;)
[20:29:20] <awallin> I wouldn't want the realtime kernel, just simulator-emc
[20:29:43] <mozmck> I don't think I built the simulator.
[20:30:51] <mozmck> I've been running a rtai patched kernel for months now on 9.04 and now on 9.10 with no problems. This is on my main computer.
[20:32:06] <awallin> if I ever get that cutting-simulator going which I'm planning then it would be interesting to drive it with the emc2 interpreter
[20:32:32] <mozmck> what will it do?
[20:32:49] <awallin> 'backplot' or 'verify' for CAM toolpaths
[20:34:03] <mozmck> doesn't emc have that? what would yours be used for?
[20:34:25] <awallin> backplot yes. That's the path the tool will take
[20:34:37] <awallin> but verify or cutting simulation models the stock also
[20:35:02] <awallin> at first it would be just visualization and verification of the CAM-code
[20:35:14] <awallin> later it could maybe be used for adaptive toolpaths
[20:36:25] <mozmck> Oh, I see. I've seen adaptive toolpaths mentioned, I think on freesteel.co.uk
[20:37:02] <awallin> http://www.youtube.com/watch?v=T3r9pFKh9lI
[20:37:12] <mozmck> how would the visualization help with that?
[20:37:17] <awallin> the freesteel code is sold as a HSM addon to mastercam by Cimco
[20:38:05] <awallin> the adaptive code suggests a path for the cutter to take. then you look from the simulation how much material that move would remove. if it's ok then we execute that move. if it removes too much (or little) we need to change the move
[20:38:31] <mozmck> doesn't heekscam have adaptive code? I was thinking it came from freesteel...
[20:38:43] <mozmck> I see, so you can look at the output and tweak it.
[20:38:51] <awallin> yes, heeks has some early version of the freesteel work
[20:39:30] <awallin> here's a very naive "pixel-mower" version http://www.youtube.com/watch?v=-IyN1wQDE0E
[20:39:58] <awallin> the algorithm doesn't know anything about the shape of the part. it just follows orders to try and cut X amount on each move
[20:41:30] <mozmck> so the simulator would be something like the one in the first video you just posted.
[20:42:09] <awallin> yes, right now my idea is to use an octree data structure. that allows for undercuts as well as 5-axis machining.
[20:42:45] <awallin> there was some work on a z-buffer model earlier, but that is only for 3-axis machining and probably has resolution/scalability problems
[20:43:02] <awallin> it was called gdepth, and jepler made it work very quickly (as usual)
[20:43:21] <mozmck> could it be used as a backplot module in emc?
[20:43:45] <awallin> yes, or call it verify to distinguish from the usual AXIS view
[20:44:00] <mozmck> that would be neat.
[20:45:12] <awallin> here is jeplers earlier work http://emergent.unpythonic.net/01169521961
[20:46:51] <mozmck> heh, I've got too much else to work on right now to dig into that too much. looks interesting though.
[20:47:24] <mozmck> bbl
[20:47:29] <awallin> bye
[22:35:39] <gtom> a simple export LANG=C in emc script resolves the locale problem, is it possible to include this in the next release???
[22:38:56] <alex_joni> doesn't that render tranlations useless?
[22:40:11] <gtom> dont think so
[22:42:29] <gtom> using gtk2 requires the use of locales, if emc does not support em we need at least a workaround
[22:42:54] <gtom> i mean i need.. :-)
[22:44:18] <alex_joni> hmm.. when I want to start emc2 translated I usually do an export LANG=de, then run the runscript
[22:44:37] <alex_joni> that way the config picker, and the GUI (if it supports i18n) get translated
[22:44:50] <alex_joni> doing an export LANG=C will probably override that
[22:45:11] <gtom> this works with emc & python, not with a gui that uses locale
[22:46:03] <gtom> not to think about the future (widestrings etc) that require the use of locale
[22:47:42] <gtom> another way to get around the problem will be to change the STRTOD in interp_read
[22:48:39] <alex_joni> gtom: I'm sure that falls under the remark from jepler (patches to fix it will be considered ;)
[22:49:38] <gtom> hope so, in the meantime i'll do a export LANG=C in my project to avoid the STRTOD- Error... :-)
[22:50:27] <gtom> got the preview working... :-)
[22:50:38] <alex_joni> hmm.. shouldn't you only need to export something related to NUMERIC?
[22:50:45] <alex_joni> LC_NUMERIC or whatever it's called?
[22:50:49] <alex_joni> gtom: saw that, nice
[22:52:16] <gtom> how can i do that??? export LC_NUMERIC=en_EN ???
[22:53:11] <gtom> http://www.cncecke.de/forum/attachment.php?attachmentid=12694&d=1258191147
[22:53:42] <gtom> made some "new" stuff also...
[22:54:25] <gtom> like: "smooth jog", jogging with acceleration... starting slow, getting faster... :-)
[22:54:57] <gtom> or the new Editor: you can see the results in the preview as you enter NC code...
[22:55:45] <alex_joni> gtom: I can see that, as I am registered
[22:55:52] <alex_joni> but I am pretty sure noone else can in here
[22:56:03] <alex_joni> it's best to put it on imagebin.org or similar
[22:56:23] <gtom> forgot that, sorry...
[22:59:13] <Jymmm> gtom: It's best that you just turn in your calipers and leave for such an atrocity and nether you, you your children's children, to ever return again!
[23:02:04] <gtom> ???
[23:02:06] <alex_joni> Jymmm: not funny ;)
[23:02:13] <alex_joni> he might take you seriously
[23:02:17] <alex_joni> gtom: ignore him :)
[23:02:30] <gtom> sometimes i do
[23:02:38] <gtom> :-)
[23:23:10] <Jymmm> alex_joni: Then why were you winking at me?