IRC log for #brlcad on 20111026

00:02.40*** part/#brlcad n_reed (~molto_cre@BZ.BZFLAG.BZ)
03:34.37*** join/#brlcad DarkCalf (DC@173.231.40.98)
04:02.33CIA-48BRL-CAD: 03brlcad * r47325 10/brlcad/trunk/ (NEWS src/nirt/interact.c):
04:02.33CIA-48BRL-CAD: add support for carriage returns in addition to newlines since nirt is doing
04:02.33CIA-48BRL-CAD: it's own parsing thing. untested, but should just make windows-style text files
04:02.33CIA-48BRL-CAD: act like they have double newlines (and should make ancient mac-style text files
04:02.33CIA-48BRL-CAD: not look like a long single line). prompted by sf support request #3426854 from
04:02.34CIA-48BRL-CAD: Fred W (NIRT scriptfile) where he assumedly ran into this issue.
04:29.02CIA-48BRL-CAD: 03brlcad * r47326 10/brlcad/trunk/src/tclscripts/lib/Makefile.am: add new Accordian.tcl file to dist
04:30.22CIA-48BRL-CAD: 03brlcad * r47327 10/brlcad/trunk/src/tclscripts/lib/Accordian.tcl: awesome to see this widget get implemented, but it started in 2011. add footer too.
04:44.47brlcadgood luck starseeker  :)
04:52.18*** join/#brlcad IriX64 (~kvirc@bas2-sudbury98-1128565147.dsl.bell.ca)
15:49.36*** join/#brlcad ibot (~ibot@rikers.org)
15:49.36*** topic/#brlcad is BRL-CAD Open Source Solid Modeling || http://brlcad.org || http://sf.net/projects/brlcad || #brlcad logs: http://ibot.rikers.org/%23brlcad/ || BRL-CAD release 7.20.2 is posted (20110701) || BRL-CAD is participating in the ESA Summer of Code in Space!
15:53.32*** join/#brlcad ibot (~ibot@rikers.org)
15:53.32*** topic/#brlcad is BRL-CAD Open Source Solid Modeling || http://brlcad.org || http://sf.net/projects/brlcad || #brlcad logs: http://ibot.rikers.org/%23brlcad/ || BRL-CAD release 7.20.2 is posted (20110701) || BRL-CAD is participating in the ESA Summer of Code in Space!
16:42.31*** join/#brlcad n_reed (~molto_cre@BZ.BZFLAG.BZ)
18:40.23*** join/#brlcad abhi2011 (~chatzilla@ip170-79-211-87.adsl2.static.versatel.nl)
21:55.27*** part/#brlcad n_reed (~molto_cre@BZ.BZFLAG.BZ)
22:45.56CIA-48BRL-CAD: 03brlcad * r47334 10/brlcad/trunk/NEWS:
22:45.56CIA-48BRL-CAD: bob added radio buttom options to archer for selecting among various lighting
22:45.56CIA-48BRL-CAD: model styles particularly relevant for visualization of BoT models. the modes
22:45.56CIA-48BRL-CAD: allow a user to specify double-sided polygon lighting so that even (incorrect)
22:45.56CIA-48BRL-CAD: back-facing polygons will be visible.
22:52.24brlcadabhi2011: what was that google docs link?
22:53.30abhi2011brlcad: https://docs.google.com/drawings/d/1Fp44t9AcZZurbNufXti64u-23CRjC49Z3r0WSmim5tU/edit
22:55.53brlcadthx
22:56.00brlcadabhi2011: presume you've read http://www.bulletphysics.org/mediawiki-1.5.8/index.php?title=Collision_Callbacks_and_Triggers yes?
22:57.30abhi2011brlcad: yes I am aware of the callbacks and I use them to calculate the force at the right point, before the simulation step
22:57.52abhi2011they do not however explain what to do , so as to insert contact points correctly in the pipeline
22:57.59brlcadabhi2011: the approach sounds good -- the only part I'm concerned about is whether manually calculating and applying forces yourself is necessary
22:58.14abhi2011it probably is not
22:58.38abhi2011however as a first shot we can try that
22:58.49brlcadapplying forces won't take care of torque or friction .. and definitely no dynamic body behaviors
22:58.59abhi2011meanwhile i havent removed any of the code pertaining to contact points
22:59.11abhi2011yes , torque will be taken care of
22:59.22abhi2011I have to only apply the force at a distance
22:59.32abhi2011there is a API for that which I will use
22:59.48abhi2011applyforce(vector)
23:00.10abhi2011the dynamics part which actually calculates the position of the body is not touched
23:00.27abhi2011that will come into action, after I apply the forces correctly
23:00.33abhi2011so it will still work
23:00.39abhi2011friction is just another force
23:01.07abhi2011I can apply friction by calculating the tangent to the surface which is touching
23:01.27brlcadthat's my point -- YOU have to do all that work
23:01.36brlcadso it'll only be as good as all of the various forces you desribe
23:01.44brlcadthat's not at all optimal...
23:02.49brlcadnot that it's not doable -- it's that it's basically a potential subset of "real" behavior or even a subset of what bullet already calculates when left to it's own computations
23:02.58abhi2011yes
23:03.21brlcadreally sounds like we need to figure out what's up with the callbacks
23:03.36brlcadyou call setCollisionFlags() ?
23:03.40abhi2011yes
23:03.55abhi2011the callbacks allow me to examine the contact points
23:03.59abhi2011but not insert them
23:04.53brlcadI don't see any reference to CF_CUSTOM_MATERIAL_CALLBACK in existing code
23:05.07brlcadthat seemed critical to several examples
23:05.59brlcadhm, I don't see setCollisionFlags() either for that matter..
23:06.05abhi2011yes, as when I last checked that, it was to set a custom friction specific to the material
23:06.06brlcadwhere is it being called?
23:06.27abhi2011yes I had tried that, it is used in run_physics()
23:06.35abhi2011*it was used there
23:07.12abhi2011I used to set the collision flags, such that objects dont collide with each other at all
23:07.22abhi2011and then apply forces
23:07.35brlcadthe behavior you described earlier where you generated the contact manifolds, but they simply didn't cause a contact response really sounds related to the collision flags
23:08.01brlcadas that looks to be how bullet controls which callbacks are even called, which are used, and how updates are performed
23:08.07abhi2011well collision flags are to filter out collisions between objects
23:08.16abhi2011so if I want 2 objects to not collide
23:08.24abhi2011then I will set differnt flags for themm
23:08.36abhi2011thats what the examples do
23:08.57abhi2011yes, during collision detection the flags are ANDed
23:09.23abhi2011and if the result is false then for those 2 objectcs, there is no collision detection carried out
23:09.29abhi2011that is what the collision flags are for
23:09.45abhi2011they do not directly control the contact points
23:09.46brlcadANDed?  that'll zero it out unless it was already set
23:10.05brlcadORd would be for setting a flag
23:10.14abhi2011no by default bullet sets them all to one for all objects
23:10.27abhi2011so all objects collide with each other
23:10.40brlcadblinks
23:11.03brlcadthe "flags" represent more than just whether two objects collide
23:11.06brlcadthat's just one of the flags
23:11.37``Erik'!' != '~'
23:11.43brlcadbitwise
23:11.51brlcad| .. not &
23:12.11brlcadif you were using &, that would explain a lot
23:12.45brlcadthe wiki docs have several examples where the flags are or'd
23:12.56brlcadrefers to http://www.bulletphysics.org/mediawiki-1.5.8/index.php?title=Collision_Callbacks_and_Triggers
23:13.45``Eriksimpsons do mad men: http://www.youtube.com/watch?v=KcmM7Jh2Y3k
23:14.46abhi2011yes but I do get use the contact callbacks, and they do get called when points are added or processed
23:15.25abhi2011I get those callbacks even if I do not set mBody->setCollisionFlags(mBody->getCollisionFlags() |    btCollisionObject::CF_CUSTOM_MATERIAL_CALLBACK);
23:16.08abhi2011whenever 2 boxes overlap for example and I insert points using my own algorithm, then the contact added/contact processed callbacks etc are called
23:16.30abhi2011thse callbacks are not the places where the contact points themselves are inserted
23:16.46brlcadabhi2011: sure, but if the right flags aren't set/unset, then the values might not get used (at all or correctly)
23:17.04abhi2011thats is true
23:17.05brlcadwhat you described is that you added a bunch of contact points but they weren't getting used, right?
23:17.11abhi2011yes
23:17.24brlcadthat points almost directly at a statefulness problem
23:17.35brlcadwhich would be those silly flags
23:18.03brlcadyou could try printing out the flags for the example code that works and the flags for your code, see how they differ
23:18.07abhi2011thats is of course possible and I will try with that, however if that were the case, then bullet would not send me a contacts processed callback with the points that I inserted,
23:18.36abhi2011yes I will try with the flags once more
23:18.51brlcadright, but one of the flags seemed to be a flag that toggled whether it needed to do anything
23:19.13brlcadif that flag was unset, it'd still make all the callbacks but not perform any collision response
23:19.23abhi2011yes that is possible
23:19.32brlcadshould hunt down the documentation on what all the flags are and what they do
23:19.51abhi2011ok, yes I will have a more careful look at the flags
23:21.47brlcadcool
23:23.59``Erikbrlcad: any experience with apache mod_proxy ?
23:24.37``Erik(I wanna futz with your server, but want second eyes before pulling the trigger)
23:26.23brlcad``Erik: nope
23:27.16``Erik'k, I'll RCS up files before mods, I guess
23:27.34``Erik(can all those bazillion *~ files in the httpd conf dir be shot?)
23:29.39CIA-48BRL-CAD: 03abhi2011 * r47335 10/brlcad/trunk/src/libged/simulate/ (simphysics.cpp simrt.c simulate.c): Putting up the code which uses only forces to stabilize objects, so I can refer to this later
23:32.15brlcadthose ~ files are backup reference
23:34.16``Erikdang emacs shenanigans O.o git init && git add -a . && git commit -m 'init'
23:36.02brlcadmeh
23:36.45brlcadmy method requires zero-effort and has been sufficient for decades :P
23:40.25``Erikprovided only emacs users admin it and only one level of mistake is made... :D
23:46.25*** join/#brlcad abhi2011 (~chatzilla@ip170-79-211-87.adsl2.static.versatel.nl)

Generated by irclog2html.pl Modified by Tim Riker to work with infobot.