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)
04:02.33 CIA-48 BRL-CAD: 03brlcad * r47325 10/brlcad/trunk/ (NEWS src/nirt/interact.c):
04:02.33 CIA-48 BRL-CAD: add support for carriage returns in addition to newlines since nirt is doing
04:02.33 CIA-48 BRL-CAD: it's own parsing thing. untested, but should just make windows-style text files
04:02.33 CIA-48 BRL-CAD: act like they have double newlines (and should make ancient mac-style text files
04:02.33 CIA-48 BRL-CAD: not look like a long single line). prompted by sf support request #3426854 from
04:02.34 CIA-48 BRL-CAD: Fred W (NIRT scriptfile) where he assumedly ran into this issue.
04:29.02 CIA-48 BRL-CAD: 03brlcad * r47326 10/brlcad/trunk/src/tclscripts/lib/Makefile.am: add new Accordian.tcl file to dist
04:30.22 CIA-48 BRL-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.
04:44.47 brlcad good 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)
22:45.56 CIA-48 BRL-CAD: 03brlcad * r47334 10/brlcad/trunk/NEWS:
22:45.56 CIA-48 BRL-CAD: bob added radio buttom options to archer for selecting among various lighting
22:45.56 CIA-48 BRL-CAD: model styles particularly relevant for visualization of BoT models. the modes
22:45.56 CIA-48 BRL-CAD: allow a user to specify double-sided polygon lighting so that even (incorrect)
22:45.56 CIA-48 BRL-CAD: back-facing polygons will be visible.
22:52.24 brlcad abhi2011: what was that google docs link?
22:53.30 abhi2011 brlcad: https://docs.google.com/drawings/d/1Fp44t9AcZZurbNufXti64u-23CRjC49Z3r0WSmim5tU/edit
22:55.53 brlcad thx
22:56.00 brlcad abhi2011: presume you've read http://www.bulletphysics.org/mediawiki-1.5.8/index.php?title=Collision_Callbacks_and_Triggers yes?
22:57.30 abhi2011 brlcad: yes I am aware of the callbacks and I use them to calculate the force at the right point, before the simulation step
22:57.52 abhi2011 they do not however explain what to do , so as to insert contact points correctly in the pipeline
22:57.59 brlcad abhi2011: the approach sounds good -- the only part I'm concerned about is whether manually calculating and applying forces yourself is necessary
22:58.14 abhi2011 it probably is not
22:58.38 abhi2011 however as a first shot we can try that
22:58.49 brlcad applying forces won't take care of torque or friction .. and definitely no dynamic body behaviors
22:58.59 abhi2011 meanwhile i havent removed any of the code pertaining to contact points
22:59.11 abhi2011 yes , torque will be taken care of
22:59.22 abhi2011 I have to only apply the force at a distance
22:59.32 abhi2011 there is a API for that which I will use
22:59.48 abhi2011 applyforce(vector)
23:00.10 abhi2011 the dynamics part which actually calculates the position of the body is not touched
23:00.27 abhi2011 that will come into action, after I apply the forces correctly
23:00.33 abhi2011 so it will still work
23:00.39 abhi2011 friction is just another force
23:01.07 abhi2011 I can apply friction by calculating the tangent to the surface which is touching
23:01.27 brlcad that's my point -- YOU have to do all that work
23:01.36 brlcad so it'll only be as good as all of the various forces you desribe
23:01.44 brlcad that's not at all optimal...
23:02.49 brlcad not 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
23:02.58 abhi2011 yes
23:03.21 brlcad really sounds like we need to figure out what's up with the callbacks
23:03.36 brlcad you call setCollisionFlags() ?
23:03.40 abhi2011 yes
23:03.55 abhi2011 the callbacks allow me to examine the contact points
23:03.59 abhi2011 but not insert them
23:04.53 brlcad I don't see any reference to CF_CUSTOM_MATERIAL_CALLBACK in existing code
23:05.07 brlcad that seemed critical to several examples
23:05.59 brlcad hm, I don't see setCollisionFlags() either for that matter..
23:06.05 abhi2011 yes, as when I last checked that, it was to set a custom friction specific to the material
23:06.06 brlcad where is it being called?
23:06.27 abhi2011 yes I had tried that, it is used in run_physics()
23:06.35 abhi2011 *it was used there
23:07.12 abhi2011 I used to set the collision flags, such that objects dont collide with each other at all
23:07.22 abhi2011 and then apply forces
23:07.35 brlcad the 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
23:08.01 brlcad as that looks to be how bullet controls which callbacks are even called, which are used, and how updates are performed
23:08.07 abhi2011 well collision flags are to filter out collisions between objects
23:08.16 abhi2011 so if I want 2 objects to not collide
23:08.24 abhi2011 then I will set differnt flags for themm
23:08.36 abhi2011 thats what the examples do
23:08.57 abhi2011 yes, during collision detection the flags are ANDed
23:09.23 abhi2011 and if the result is false then for those 2 objectcs, there is no collision detection carried out
23:09.29 abhi2011 that is what the collision flags are for
23:09.45 abhi2011 they do not directly control the contact points
23:09.46 brlcad ANDed? that'll zero it out unless it was already set
23:10.05 brlcad ORd would be for setting a flag
23:10.14 abhi2011 no by default bullet sets them all to one for all objects
23:10.27 abhi2011 so all objects collide with each other
23:10.40 brlcad blinks
23:11.03 brlcad the "flags" represent more than just whether two objects collide
23:11.06 brlcad that's just one of the flags
23:11.37 ``Erik '!' != '~'
23:11.43 brlcad bitwise
23:11.51 brlcad | .. not &
23:12.11 brlcad if you were using &, that would explain a lot
23:12.45 brlcad the wiki docs have several examples where the flags are or'd
23:12.56 brlcad refers to http://www.bulletphysics.org/mediawiki-1.5.8/index.php?title=Collision_Callbacks_and_Triggers
23:13.45 ``Erik simpsons do mad men: http://www.youtube.com/watch?v=KcmM7Jh2Y3k
23:14.46 abhi2011 yes but I do get use the contact callbacks, and they do get called when points are added or processed
23:15.25 abhi2011 I get those callbacks even if I do not set mBody->setCollisionFlags(mBody->getCollisionFlags() | btCollisionObject::CF_CUSTOM_MATERIAL_CALLBACK);
23:16.08 abhi2011 whenever 2 boxes overlap for example and I insert points using my own algorithm, then the contact added/contact processed callbacks etc are called
23:16.30 abhi2011 thse callbacks are not the places where the contact points themselves are inserted
23:16.46 brlcad abhi2011: sure, but if the right flags aren't set/unset, then the values might not get used (at all or correctly)
23:17.04 abhi2011 thats is true
23:17.05 brlcad what you described is that you added a bunch of contact points but they weren't getting used, right?
23:17.11 abhi2011 yes
23:17.24 brlcad that points almost directly at a statefulness problem
23:17.35 brlcad which would be those silly flags
23:18.03 brlcad you could try printing out the flags for the example code that works and the flags for your code, see how they differ
23:18.07 abhi2011 thats 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,
23:18.36 abhi2011 yes I will try with the flags once more
23:18.51 brlcad right, but one of the flags seemed to be a flag that toggled whether it needed to do anything
23:19.13 brlcad if that flag was unset, it'd still make all the callbacks but not perform any collision response
23:19.23 abhi2011 yes that is possible
23:19.32 brlcad should hunt down the documentation on what all the flags are and what they do
23:19.51 abhi2011 ok, yes I will have a more careful look at the flags
23:21.47 brlcad cool
23:23.59 ``Erik brlcad: any experience with apache mod_proxy ?
23:24.37 ``Erik (I wanna futz with your server, but want second eyes before pulling the trigger)
23:26.23 brlcad ``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?)
23:29.39 CIA-48 BRL-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
23:32.15 brlcad those ~ files are backup reference
23:34.16 ``Erik dang emacs shenanigans O.o git init && git add -a . && git commit -m 'init'
23:36.02 brlcad meh
23:36.45 brlcad my method requires zero-effort and has been sufficient for decades :P
23:40.25 ``Erik provided 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.