#emc-devel | Logs for 2010-03-01

[03:25:41] <cradek> recursive make sucks
[03:28:56] <jepler> what are you building?
[03:29:09] <cradek> decade-old festival
[03:29:21] <jepler> ugh
[03:30:02] <cradek> for so long, big projects barely built
[03:30:06] <cradek> well you know this
[03:31:03] <cradek> "Please consider using one of the 32 headers found in section of the C++ standard."
[03:31:06] <cradek> haha
[03:31:19] <cradek> "Surely 32 headers is enough for anyone!"
[03:41:25] <jepler> cradek: maybe this "flite" is worth a look, depending how far down the road you are with festival this time around
[03:41:30] <jepler> of course, you want the scheme interpreter, don't you..
[03:41:42] <cradek> yeah I even extended it
[03:41:51] <cradek> it's actually working, just taking a while
[03:43:06] <cradek> unlike the beagle, I think this'll actually work
[09:34:56] <micges_work> jepler: I propose another fix for rdf0bdc9f4b7d: http://pastebin.ca/1816809
[09:36:07] <micges_work> it's restoring previous behavoiur of Axis: when selecting background, rotation center will be at program center
[09:39:12] <micges_work> and small remove of unused code: http://pastebin.ca/1816813
[09:51:07] <alex_joni> micges_work: it would be nice to be able to set the rotation center
[09:51:19] <alex_joni> (on some CAD programs I can do that by double-clicking middle)
[09:52:50] <micges_work> agree
[10:14:23] <micges_work> alex_joni: it is easy in 2d mode, but how to get Z coord from perpective view?
[10:18:51] <micges_work> heh it's easy too;)
[10:44:31] <alex_joni> see .. everything is easy if you know how to do it
[10:52:34] <micges_work> bbl
[12:51:33] <jepler> micges: those are both OK to commit. Looks like they are only relevant on master.
[13:40:32] <aystarik> guys, I've found scaling in a fanuc manual... G2/3 are still arcs, if R is supplied, then R is multiplied by max axis scale, if I-J-k is supplied, arc has min radius, and then linear move to the end is added.
[13:42:28] <aystarik> it is possible to supply negative scale value -- it should mirror the output
[13:45:02] <jepler> I don't understand how that treatment of arcs under anisotropic scaling is likely to give useful results. Do you?
[14:02:42] <aystarik> no...
[14:03:14] <alex_joni> that is just sooo bogus: if I-J-k is supplied, arc has min radius, and then linear move to the end is added.
[14:03:28] <alex_joni> I can't see how that can help anything
[14:03:49] <alex_joni> the arc should become an ellipse arc if scaling is not uniform..
[14:05:59] <jepler> incidentally: http://www.geometrictools.com/Documentation/ApproximateEllipse.pdf
[14:06:10] <aystarik> this is 0i manual -- quite antique... may be modern do it differently
[14:24:10] <jepler> hey skunkworks_
[14:26:26] <skunkworks_> Good morning jepler
[14:56:28] <alex_joni> jepler: http://homepage.smc.edu/kennedy_john/belipse.pdf
[15:05:56] <jepler> bresenham algorithms are not useful for cnc, because they are not constant-velocity.
[15:06:46] <jepler> they proceed at one pixel per iteration, so the velocity varies from 1 to sqrt(2)
[15:08:31] <alex_joni> well.. they are for displaying the path
[15:08:45] <alex_joni> at trajectory level one could use NURBS directly
[15:09:08] <jepler> in alternate world crazy land where we use NURBS at the trajectory level
[15:10:42] <alex_joni> in alternate world crazy land where we have non-uniform scaling for XYZABCUVW
[15:12:41] <jepler> just so long as we get arcs in all the planes, such as XV, for our troubles.
[15:16:12] <alex_joni> heh
[15:48:15] <jepler> A $375, less than a watt LED? http://mouser.com/ProductDetail/Osram-Opto-Semiconductor/OSP1G10WW2031C1000-3133-4Q9X/?qs=pD9e3CUnaqNF%252b64e83is7Q%3d%3d
[16:30:39] <aystarik> alex_joni: you might still just stretch a circle arc... All the ellipses you get with non-uniform scaling are X-Y aligned.
[16:43:14] <jepler> if we introduce ellipses we'd better be able to get them with the major and minor axes not necessarily XY aligned..
[16:44:03] <SWPadnos> and with the axes non-aligned as well
[16:46:56] <aystarik> well, then we are better to transform everything to NURBS and work there... :)
[16:49:17] <jepler> that takes a bit more than just wishing it were so.
[16:53:45] <aystarik> right... We are started with non-essential scale command, then added non-essential handling of G2/G3 in non-uniform scaling mode, and ended up with the need for handling non-axis alinged ellipses... Why stop half-way?
[17:00:33] <CIA-2> EMC: 03micges 07master * rcb6d75c06f7a 10/src/emc/usr_intf/axis/scripts/axis.py: Remove unused varables and minor redundant code
[17:00:35] <CIA-2> EMC: 03micges 07master * r8caaa7655a2d 10/lib/python/rs274/glcanon.py: Fix case when no vector was selected
[17:00:45] <jepler> micges: thanks
[17:00:49] <micges> np
[17:14:14] <micges> jepler: is it hard to attach let say mandriva build for our buildbots?
[17:14:26] <micges> (sim of course)
[17:17:11] <jepler> seb_kuzminsky: 11:14:14 <micges> jepler: is it hard to attach let say mandriva build for our buildbots?
[17:33:04] <aystarik> http://en.opensuse.org/Build_Service -- could this one put to use?
[18:05:00] <cradek> at the risk of sounding like a broken record, all layers past interp/canon handle arbitrary arcs, and in fact you can currently get some by rotating XY and making a G18/G19 arc.
[22:08:10] <ries_> ries_ is now known as ries