#emc-devel | Logs for 2008-01-07

[12:44:55] <fenn> what would be involved in porting AXIS to windows? is this feasible?
[12:47:24] <SWPadnos> it's python, so it should be mostly ported already
[12:47:32] <SWPadnos> assuming the underlying libraries are there
[12:47:56] <SWPadnos> but then there's "gcodemodule.cc" (and maybe others) that would need to be ported
[12:47:58] <fenn> that's what i thought, but it seemed too easy
[12:48:04] <SWPadnos> ^^
[12:48:40] <SWPadnos> so the code that makes the screen look all nice should be pretty easy (if you ignore the GL part of it)
[12:48:52] <SWPadnos> but the rest isn't so easy
[12:49:03] <fenn> windows can do opengl though right?
[12:49:06] <alex_joni> I tried once
[12:49:07] <SWPadnos> (but this is just me thinking about it, I'm far far from a Python / AXIS expert)
[12:49:18] <SWPadnos> yes, Windows has OGL
[12:49:24] <alex_joni> fenn: you'll have a hard time to compile all parts
[12:50:01] <SWPadnos> hmmm. thinking about the UI, I don't know how much of AXIS is X-centric
[12:50:17] <SWPadnos> I know the author is Xcentric though :)
[12:50:20] <alex_joni> SWPadnos: that's not a very big problem :)
[12:50:36] <alex_joni> but the issue is compiling some underlying stuff
[12:50:46] <alex_joni> I would start by removing all HAL traces..
[12:50:57] <SWPadnos> right, like gcodemodule (is togl also a .cc?)
[12:51:02] <alex_joni> yes
[12:51:04] <fenn> i think its kinda sloppy to have axis interact with hal anyway
[12:51:12] <alex_joni> I know I had problems with togl back then
[12:56:28] <SWPadnos> fenn, there's some AXIS_NOHAL environent var you can set to prevent AXIS from creating HAL pins
[12:56:39] <SWPadnos> I don't recall the name, but it was on the list lately
[12:58:21] <fenn> what are the axisui.jog pins for anyway?
[12:58:41] <SWPadnos> dunno
[13:03:34] <alex_joni> fenn: they select which axis is activated
[13:03:45] <alex_joni> so if you select axis X on the GUI, the forst jog pin is high
[13:03:53] <alex_joni> (that helps when you have a handwheel)
[13:04:21] <SWPadnos> ah - they're "jog-axis-select-output" pins
[13:04:52] <fenn> hmm.
[13:04:52] <alex_joni> right
[13:04:59] <alex_joni> you can easily do without
[13:05:12] <alex_joni> fenn: I recently added a AXIS_NO_HAL env support
[13:05:19] <alex_joni> so AXIS runs on systems without HAL
[13:05:21] <fenn> i think it should go away entirely
[13:05:50] <fenn> er, the hal stuff in any gui i mean
[13:06:12] <fenn> this could be done with nml messages to select the handwheel axis
[13:06:37] <fenn> then halui can export that
[13:07:32] <fenn> and it could be a number instead of having 9 pins
[13:08:21] <alex_joni> fenn: that may be so
[13:17:50] <steve_stallings> steve_stallings is now known as steves_logging
[13:26:18] <alex_joni> fenn: you're looking at compiling axis?
[13:27:12] <fenn> yes, where should i start?
[13:27:26] <fenn> looking that is
[13:27:32] <fenn> i've never compiled anything on windows
[13:27:35] <alex_joni> hmm.. I reinstalled about twice since then, and I don't have much left
[13:27:45] <alex_joni> first you gotta figure out what compiler you'll use
[13:27:56] <fenn> can i use gcc?
[13:27:57] <alex_joni> iirc I could only use some M$ shit for the python-gl stuff
[13:28:07] <alex_joni> fenn: only if you use mingw I think
[13:28:28] <alex_joni> but you need to compile parts of emc first
[13:28:40] <alex_joni> although I think it will be slightly easier to start with rcslib
[13:28:59] <fenn> i thought that already worked with windows
[13:29:02] <alex_joni> (take an old rcslib from CVS, that one has projects for M$ Vcc)
[13:29:20] <alex_joni> you then need to take a recent emc.hh and emc_nml.hh, and whatever has changed
[13:30:08] <alex_joni> I think that's the quickest way to getting the needed infrastructure for AXIS to run..
[13:30:37] <alex_joni> the other way is installing cygwin and cygwin-X and trying there
[13:30:50] <fenn> when you ran tkemc on windows did you use cygwin?
[13:31:01] <alex_joni> no, tkemc runs natively under tcl/tk
[13:31:33] <fenn> hmm.. so you're just telling me how to set up the compiler
[13:32:26] <alex_joni> so far .. yes
[13:32:47] <alex_joni> basicly you have 2 choices
[13:32:55] <alex_joni> port stuff over to win
[13:33:02] <alex_joni> or try to get it running under cygwin
[13:33:28] <alex_joni> when I tried back then, there were a couple of things that didn't work under cygwin like python-gl iirc
[13:34:01] <alex_joni> it might have changed in the last 2? years
[13:34:16] <fenn> probably
[13:34:33] <fenn> but axis still uses weird python-gl stuff
[13:34:43] <fenn> i.e. not a standard python module
[13:35:04] <alex_joni> http://cygwin.com/ml/cygwin/2004-10/msg00677.html
[13:35:23] <alex_joni> seems other people are running python-gl stuff in cygwin
[13:35:28] <alex_joni> so that's good
[13:36:56] <fenn> ok well, this is far down on the priority list, so maybe the win-empire will crash and burn before i get around to it
[13:37:16] <alex_joni> ha
[13:41:33] <alex_joni> fenn: http://eneas.juve.ro/~juve/emc/emc_running_win_xp.JPG
[13:42:24] <alex_joni> that was with emc1, and existing projects for M$ Vcc
[13:42:33] <lerneaen_hydra> alex_joni; that's just so wrong ;)
[13:42:35] <alex_joni> it still took me about a week till I got it working
[13:42:47] <alex_joni> lerneaen_hydra: I was young and naive :P
[13:43:37] <lerneaen_hydra> It would seem that you've learned from your endeavor :p
[13:43:53] <alex_joni> well.. I'm not so sure I would try again :D
[13:44:05] <alex_joni> AXIS runs in ssh -x under CygwinX just fine..
[13:44:26] <lerneaen_hydra> that sounds a lot simpler
[13:44:46] <lerneaen_hydra> IIRC there are even dedicated x servers for windows
[13:44:53] <lerneaen_hydra> though their name seems to elude me
[13:45:48] <alex_joni> Xming
[13:46:01] <lerneaen_hydra> ah yes that's right
[13:46:14] <lerneaen_hydra> I could only remember exceed which is not Free
[13:46:23] <alex_joni> http://sourceforge.net/projects/xming
[13:46:24] <lerneaen_hydra> in both meaning
[13:46:27] <lerneaen_hydra> meanings
[13:46:28] <SWPadnos> X-win32
[13:46:33] <SWPadnos> also not free or Free
[13:46:42] <alex_joni> yeah.. I remember that one
[13:47:11] <SWPadnos> there was another one too, though I forgot the name. their claim to fame was that they had accelerated Parhelia drivers for Linux
[13:47:22] <alex_joni> hmm.. Xming-portable-PuTTY sounds good :D
[13:47:41] <SWPadnos> Xi Graphics, that was it
[13:47:45] <alex_joni> there is a Zming-mesa.. so that might provide accel
[13:48:09] <SWPadnos> oh, but not for Windows. how nice!
[13:48:19] <alex_joni> I meant Xming-mesa
[14:00:05] <jepler> I jettisoned the 'pyopengl' package, so you don't have to worry about getting it built for mingw. You do have to worry about getting the python and tk extensions included in emc built, that includes an OpenGL API and a port of the tk "ToGL" widget.
[14:00:30] <jepler> (at the time I started axis, bdi shipped a buggy pyopengl package; its ToGL just crashed when you attempted to use it)
[14:00:46] <jepler> (I struggled with it for awhile then gave up and just put what I needed from it -- a tiny subset -- in emc)
[14:02:41] <alex_joni> emc extension does mean some parts of libnml though
[21:36:13] <cradek> it's interesting to note that if we have joint constraints we could more easily have anisotropic scaling
[21:36:34] <skunkworks> * skunkworks needs to look that word up..
[21:37:30] <cradek> scaling where the different directions don't have to all be the same
[21:39:01] <cradek> it would turn circles into ellipses for instance
[21:41:15] <skunkworks> ah - so - it might work for dia/rad on lathe?
[21:42:38] <jepler> hm, keystick doesn't run for me ..
[21:42:38] <jepler> emc/usr_intf/keystick.cc 1392: emcStatus buffer not available
[21:43:07] <cradek> skunkworks: good question, I hadn't thought of that application.
[21:43:12] <jepler> ugh it works if I add a 'sleep' to the emc script
[21:43:19] <cradek> I think that still depends on interpreter hackery though
[21:43:50] <cradek> I think for diameter mode, IJ in arcs are radius but they end up twice as big, or some other crazy-ass thing
[21:44:16] <SWPadnos> the trouble with scaling is offset
[21:44:56] <SWPadnos> setting the Z scale to 0.5 would make R act like D on a lathe (I think), but relative to what?
[21:45:18] <cradek> thanks for reminding me not to worry about this
[21:45:20] <SWPadnos> you'd have to be at R=D=Z=0 for that to not cause following error badness
[21:45:22] <SWPadnos> heh
[21:45:51] <cradek> too many actually-useful things to worry about
[21:45:58] <SWPadnos> true enough
[21:47:07] <skunkworks> heh
[21:47:14] <skunkworks> sorry about that..
[21:47:33] <skunkworks> * skunkworks doesn't really care either.
[21:57:04] <CIA-20> EMC: 03jepler 07TRUNK * 10emc2/scripts/emc.in: (log message trimmed)
[21:57:04] <CIA-20> EMC: * don't run graphical programs (e.g., @WISH@, popimage) when DISPLAY is unset
[21:57:04] <CIA-20> EMC: * don't use xterm to start keystick when DISPLAY is unset
[21:57:04] <CIA-20> EMC: * I found that without an extra xterm startup time, keystick did not
[21:57:04] <CIA-20> EMC: consistently start for me, so I added a 'sleep'. Someone should look
[21:57:07] <CIA-20> EMC: into the question of how to make sure that emcsvr is ready before starting
[21:57:09] <CIA-20> EMC: other processes, or else fix keystick to wait a short time for the stat