#bitfighter IRC Log

Index Search ←Prev date Next date→

IRC Log for 2014-08-22

Timestamps are in GMT/BST.

02:31:45Akien has joined
02:52:23Akien Quit (Remote host closed the connection)
05:17:42Darrel Quit (Ping timeout: 272 seconds)
05:18:03Darrel has joined
05:42:40LordDVG has joined
05:46:37Darrel Quit (Ping timeout: 260 seconds)
05:47:59Darrel has joined
07:06:26Nothing_Mucho.o
08:27:42LordDVG Quit (Ping timeout: 272 seconds)
08:59:20LordDVG has joined
09:38:36YoshiSmb_m has joined
09:39:16YoshiSmb_mhi guys
09:39:38YoshiSmb_mkaen are you there?
09:42:39Nothing_MuchHey YoshiSmb_m
09:42:47Nothing_MuchI think kaen's been super busy
09:42:51Nothing_MuchHe hasn't responded to me once on Skype :(
09:43:01YoshiSmb_mhey
09:43:26YoshiSmb_mwow
09:43:29YoshiSmb_mdo you play nethack?
09:44:04Nothing_Muchnope
09:44:11Nothing_MuchI'm not that good at text only games :(
09:44:20YoshiSmb_mand you should nm, you should...
09:45:15YoshiSmb_mbut you can use the Native Graphical Interface?
09:45:44YoshiSmb_min SlashEN
09:45:46YoshiSmb_m*EM
09:46:11YoshiSmb_mi got 2 wish, after starting the game
09:46:44YoshiSmb_mit's like i got lucky
09:49:48YoshiSmb_m16:22:44 Nothing_Much I didn't see
09:50:12YoshiSmb_msorry if i dint put it well
09:50:47YoshiSmb_mi just copy it using my mobile (kinda hard)
09:52:14Nothing_Muchoh man
09:52:35Nothing_Muchwell, idk, I need some form of graphics to help me distinguish what I'm looking at
09:52:50YoshiSmb_myea
09:54:31YoshiSmb_mwhat you say early, was very funny btw
09:59:48Nothing_Muchlol
09:59:49Nothing_Much>.>
10:00:08Nothing_MuchI recall that 019b was still the only one there
10:00:15Nothing_MuchBut then BF got hacked
10:00:27Nothing_Much>.>:;
10:00:40YoshiSmb_magain...
10:00:43YoshiSmb_m?
10:01:16Nothing_Muchoh no
10:01:32Nothing_Muchthat was when I only saw 019b
10:01:41Nothing_Muchafter the hack, that's when I saw 019c
10:01:42YoshiSmb_mah
10:01:48Nothing_Muchso that's a good update
10:03:43YoshiSmb_myep
10:03:56YoshiSmb_mwww.srb2.org
10:05:13Nothing_Muchwow, that looks pretty bad :(
10:05:34YoshiSmb_mwhat?
10:08:06Nothing_Muchit looks not that good, honestly
10:09:18YoshiSmb_mwell
10:09:30YoshiSmb_mat least you could try
10:15:29Nothing_MuchI could
10:15:32Nothing_MuchLemme see if it's on Linux
10:16:57Nothing_MuchNo GNU/Linux support :(
10:17:02YoshiSmb_mcheck SRB2 Ports
10:17:21YoshiSmb_mon download section
10:22:34YoshiSmb_mbut sorry if you couln't use it :-(
10:35:31YoshiSmb_m Quit (Remote host closed the connection)
10:55:09Akien has joined
11:13:12YoshiSmb_m has joined
11:21:12starseeker has joined
11:21:36starseekerhi! is this where the bitfighter devs are, or is this a user/gamer channel?
11:32:09YoshiSmb_mboth
11:32:30starseekercool. I came across bitfighter while reading up on poly2try and clipper
11:32:39starseekersorry, poly2tri
11:33:00starseekerit looks like some folks with bitfighter have done some fairly extensive work to combine the two
11:35:46starseekerwas curious to chat with the folks who had done that work, but don't want to annoy the gamers ;-) Is the technical discussion section in the forum where such conversations normally occur?
11:39:10YoshiSmb_myes
11:39:57YoshiSmb_mtake in mind, everyone is busy or not (doing homework r he's afk).
11:40:03YoshiSmb_m*or
11:41:45YoshiSmb_mi almost forgot, Welcome Starseeker
11:46:16starseekersure :-)
11:46:37starseekerthank you
11:47:02starseeker came looking to discuss triangulation, but now may end up trying bitfighter tonight :-)
11:49:43YoshiSmb_mYou're Welcome :-)
12:24:43raptor has joined
12:24:43ChanServ sets mode +o
12:24:57raptorgood day!
12:25:14YoshiSmb_mhi raptor
12:25:21raptorhello
12:25:37raptorstarseeker: I am one of the devs that worked with poly2tri quite a bit
12:25:46raptorif you have questions
12:26:00raptoralso, welcome!
12:48:59YoshiSmb_m Quit (Remote host closed the connection)
13:09:22Nothing_Muchhey, a new guy!
13:09:24Nothing_Muchwoo!
13:09:28Nothing_Muchstarseeker: welcome!
13:09:43Nothing_Much(I'm a gamer though, so don't pay attention to me :P)
13:18:20YoshiSmb_m has joined
13:31:07starseekerNothing_Much: thanks :-)
13:32:06starseekerraptor: I read various comments on the poly2tri website and stackoverflow, and from the sound of things you guys have had pretty good luck with using clipper to "clean up" input before feeding it to poly2tri
13:32:31raptoryes
13:32:34raptorwe love clipper
13:32:49raptorI think it's one of the best libraries I've ever worked with
13:33:46starseekerI am interested in trying something similar, but the projects I work with can't use GPL code (long story). As near as I can tell with a casual search, it looks like the bits in bitfighter that pipe data from bitfighter->clipper->poly2tri->bitfighter are part of the GPL codebase?
13:34:19raptoryes
13:34:38raptorwe have several GeomUtils methods that call clipper
13:34:51starseekerI was wonder if a) those pieces are generic enough that they might be of wider interest to the computer graphics community as a whole, and b) whether the folks who did that work might consider licensing just those bits under a license like that of clipper or poly2tri
13:35:07raptorhmmm
13:35:26starseekernot proposing re-liecensing all of bitfighter, of course
13:35:44raptorwell, i probably did most of the latest work in getting clipper to use poly2tri
13:35:55starseekerit's cool if the answer is no, but I figured it was better to inquire before going down the road of do-it-yourself
13:36:10raptorand the other devs that helped out are pretty cool with the code... especially since you asked nicely :)
13:36:17raptorso I would say take whatever you want
13:36:33raptoralthough some it is hightly specific to bitfighter functions, etc.
13:36:41starseeker nods
13:37:22starseekerwhat I was thinking was to identify the bits that might be generally applicatable (I'm guessing the floating point data -> clipper, clipper->poly2tri, and (maybe) poly2tri back out)
13:37:40raptorso for a) I think the methods could easily be made more generic, but probably still tied to c++
13:37:50starseeker nods - C++ is cool
13:38:23raptorI'm not sure how much you've looked at the code, but the workflow is for generating 2D robot navigation zones
13:38:36raptorand it goes like this:
13:39:58raptorfloating point data -> clipper -> more custom cleaning for really wacky use cases -> poly2tri -> Recast (customized for 2D) -> floating point data
13:41:11starseeker nods - that definitely sounds like the general problem most people are hoping to solve with poly2tri
13:41:28starseekerheh - is this you? http://www.gamedev.net/page/resources/_/technical/artificial-intelligence/generating-2d-navmeshes-r3393
13:43:17starseekerah, here's the stackoverflow link: https://code.google.com/p/bitfighter/source/browse/zap/GeomUtils.cpp?spec=svnb9286436e08332dd50d7302bb111a6cf6654ad1a&r=b9286436e08332dd50d7302bb111a6cf6654ad1a#1046
13:45:53raptortaht first link is not me... but I admit that we probably did most of the pioneering work in getting these libraries to work together
13:46:28starseeker nods
13:49:01starseekerraptor: I guess what I should do is put together the pieces that are needed for a library under an API, and then post the result in the forums to make sure all the devs with a hand in it are OK with the results?
13:50:33raptorActually, you can use whatever you want with permission
13:50:51raptorwe use a custom container 'TNL::Vector'
13:50:59raptorwhich you'll have to change to std::vector
13:51:22raptorand a Point object
13:52:01starseeker nods
13:52:21raptori mean use whatever you wish with our full permission :)
13:53:03starseekerraptor: cool - thank you!
13:54:47starseekerraptor: I just want to be sure (part politeness, part lawyer vaccination) to do my homework and make sure everyone's cool. Y'all have done some really cool work here
13:55:17starseeker was also favorably impressed with fontstash
13:56:33starseeker gets the sense that robust triangulation of polygons is a *very* common problem in the gaming world...
13:56:58starseekerraptor: do you guys convert poly2tri output into triangle strips, or is that not necessary for your application?
14:01:42raptorhi again, sorry I'm at work, and people keep coming into my office...
14:02:05raptorstarseeker: what do you mean by 'triangle strips'?
14:02:06starseekerraptor: no problem - I'm used to IRC conversations :-)
14:02:28starseekerhttp://users.telenet.be/tfautre/softdev/tristripper/
14:02:50starseekerraptor: if you'd rather I can return later tonight - don't mean to bother you at work
14:03:28starseekertriangle strips (as I understand the term) are used to more compactly represent triangles for display
14:03:37raptorah i see
14:03:39raptorno
14:03:49raptorwe use a custom version of RecastNavigation
14:04:02raptorit merges all the triangles into efficient convex polygons
14:04:14starseeker nods - that makes sense
14:04:26raptorbecause convex polygons are what's needed for good bot nav zones
14:05:13starseekerso you're actually using them for geometry interaction - right, strips wouldn't make sense for that (rather the contrary)
14:05:53raptorthe output polygons are used for pathfinding
14:07:56raptortechnically, the triangles themselves could be used for pathfinding since they're guaranteed to be convex
14:08:17raptorbut that adds huge overhead for pathfinding with having so many zones
14:08:26starseeker nods
14:08:40starseekerhadn't heard of recastnavigation before - neat stuff
14:10:56raptorso triangle stripping would probably be good in rendering situations, but not at all for bot navigation
14:11:42starseeker nods - looks like the rendering is vector graphics based on the screenshots, so it was a silly question - my bad
14:12:30raptorno not a silly question at all
14:12:54raptorat first glance no one would know how we use these at all without extensive code searching :)
14:13:48starseekerhow old is bitfighter? I don't remember it from my early Linux exploration days
14:14:00raptorhmmm... maybe 6 years or so?
14:14:10Nothing_Muchstarseeker: It used to be known as Zap! and ZAP
14:14:20Nothing_MuchBitfighter was made in 2010, right?
14:14:21raptorwatusimoto is the head developer who started it - he'll be back in a couple days, i think
14:14:32starseekerah, yeah - that would put it after my "exploring linux gaming" phase
14:14:53Nothing_MuchZap! was released in 2004 and was only on Windows and Mac, sadly :(
14:15:09raptorone year we'll get into debian...
14:15:34starseekerheh - getting into Debian is like climbing Mount Everest for software
14:15:40Nothing_Muchraptor: I would've done it, but kaen said he'd do it because Debian has extremely high standards
14:16:01Nothing_Muchand I probably wouldn't be the best.. person to get BF on Debian
14:16:12raptorwe got so close the last time kaen put effort into it - i think the issue was a bug in a debian font package
14:17:03starseekerdo they make you use debian packages for clipper/poly2tri/recastnavigation/fontstash/etc.?
14:17:12raptoroh yeah
14:17:12Nothing_Muchhttps://bugs.debian.org/cgi-bin/bugreport.cgi?bug=705303
14:17:15Nothing_MuchA response!!
14:17:19Nothing_MuchAutomated though
14:17:24Nothing_MuchBut a response nonetheless
14:18:00raptorha!
14:19:59starseekerwe had to add some extensive CMake logic to manage build with both system and bundled versions of libraries based on either user settings or system detection results
14:20:20kaenhi all
14:20:24starseekersomeday I'm going to polish that up and try to get the CMake guys to stick it in the standard macro package...
14:20:49kaenerm, yes... the debian package
14:20:58Nothing_MuchHey kaen, how've you been?
14:20:58raptorkaen! welcome back!
14:21:14kaenI halfway ran out of steam and halfway got abducted by people giving me money to keep their websites functioning
14:21:25raptorhaha
14:21:27kaenI've been very well :)
14:21:35raptorhow is the money front going?
14:21:38Nothing_MuchAwesome dude!
14:21:50kaenmuch better than it ever has :D
14:21:57raptoroh great!
14:21:59kaenI got hired on full-time by one of my contracting clients
14:22:10kaenI've got a salary and benefits now, like a big boy \o/
14:22:21Nothing_Muchyou're so lucky kaen
14:22:24raptorand the money arrives on time and everything?
14:22:29kaenyes indeed
14:22:44kaento both
14:23:20kaenI think the last roadblock for the debian package is the font package problem
14:23:21raptorexcellent
14:23:30raptorI was thinking about the font problem -
14:23:43raptori think we should just use the system font, ugly and all
14:23:43Nothing_MuchDude, there's an automated message regarding the lack of activity in that bug report
14:23:49raptorand that might get them to fix it faster
14:23:55Nothing_MuchDoes that mean you have to go through the process all over again? :(
14:23:56kaenI don't remember exactly, but iirc there should be a (functioning?) script to build a n official debian package
14:24:05kaenoh, no
14:24:09kaenI just have to resubmit it
14:24:15raptorwe have a debian branch, i think
14:24:30kaenyeah, the original "intent to package" report was opened like two years ago
14:24:32raptoroff of 019c
14:24:37kaenso I that autoresponse isn't surprising
14:24:39kaenah, that's right
14:24:45raptor019c might be around for a while...
14:24:48kaenerm, we're still on 019c right?
14:24:49kaenok
14:24:57Nothing_MuchOh by the way, there's been a hack too
14:24:57kaenworried I had missed a release :P
14:25:03Nothing_Muchsadly
14:25:07kaenI caught that :/
14:25:08raptor020 is broke and I cannot even begin to guess what all the problems might be...
14:25:14kaenoh no
14:25:21raptori mean, it runs...
14:25:26kaenhmm
14:25:32raptorthere's at least 2 major refactors going on right now
14:25:37kaenI guess the refactoring got more involved
14:25:39kaenah, yep
14:26:00raptorone with game settings, one with Level parsing/loading
14:26:08kaenyuck
14:26:20raptorsam686 will show up every once in a while and fix our crashes we introduced...
14:26:33kaenheh
14:27:12raptoralso, with windows 9 around the corner, and the possibility that 32bit support will be dropped, we'll probably have to do a win64 port
14:27:28kaenI see
14:27:43kaenshould just be a matter of recompiling the dll's right?
14:27:44raptorwhich actually isn't so bad, except for the libraries
14:27:48raptoryes
14:28:23raptoroh, and all this is happening at a snail's pace - we've mostly been vacationing and taking it easy for the summer
14:28:36kaengood :)
14:30:07kaenI've been doing some ninja advertising
14:30:10kaenmostly on reddit
14:30:14raptorninja?
14:30:22kaentaking out ads without telling you guys :P
14:30:26raptorohhh, haha
14:30:53kaenI feel bad for leaving you hanging :/
14:30:55Nothing_Muchby taking out you mean putting on ads on facebook and stuff?
14:30:57raptornah...
14:30:59raptoryou didn't really
14:31:06raptoryou gave us notice :)
14:31:31raptorand the summer happened anyways
14:38:41kaeninteresting. I didn't realize we were breaking ground with poly2tri
14:38:45raptoroooo... freeorion is going to release soon
14:39:16raptorkaen: oh yeah, we totally pioneered with it
14:39:28kaenI have definitely noticed a lot of people interested in automatic bot navigation
14:39:34kaenI see it on reddit pretty frequently
14:40:08raptori had several discussions with the head developer and got him to suggest clipper
14:40:35kaenreally cool
14:40:58raptorthe main issue is that poly2tri is free, super fast, but the algorithm itself doesn't allow for certain robustness
14:41:15raptorand there isn't really a fast alternative with such liberal licensing
14:41:19kaenyes
14:42:32raptoractually, starseeker may have a good idea - maybe I should just pull out our code and make a simple library with our code that combines clipper/polytri
14:42:45raptorpost it on github in the public domain
14:43:01starseekerraptor: that would be even better - someone who knows what they are doing :-)
14:43:24starseekerif it gets enough momentum from lots of projects, everyone can benefit from robustness improvements
14:43:37kaeninteresting
14:43:51starseekerkinda what's happened with clipper itself, actually
14:44:47raptorclipper is just amazing
14:45:17raptorevery single time i thought clipper was doing something wrong... it turned out to be me instead
14:47:49kaenit also outperforms boot::polygon and boost::geom
14:47:55kaenwhich is still baffling to me
14:48:21raptoryeah, me too
14:49:07kaenI talked to some guys who use it for PCB layouts and trace routing
14:49:17kaenthey said it's meant for really large N
14:50:59raptorboost:geom?
14:51:09raptor::geom
14:51:36LordDVG Quit (Remote host closed the connection)
14:52:13kaenyes, where the more complicated polygon operations live
14:52:23kaentemplatized beyond usability, of course
14:52:50starseekernot to mention requiring all of boost to be there... always fun on Windows machines
14:53:19kaenI think those two are header-only, although I don't remember anyway
14:53:29kaenwe actually have header-only stuff in bf already
14:53:47kaenbut, at any rate, clipper beats it from every angle
14:54:06starseekereven boost headers (in my experience) tend to require a lot of other headers... best use of a self contained boost subset I know of is the gecode constraint solving library's floating point support
14:54:55starseeker nods - if clipper hasn't won some kind of free software award, it's because someone isn't paying attention
14:55:50kaenstarseeker: what's your project?
14:55:55kaenis it a game as well?
14:56:03starseekerkaen: nope, cad - BRL-CAD
14:56:11kaenawesome!
14:56:16raptori don't know what that is...
14:56:25YoshiSmb_m Quit (Remote host closed the connection)
14:56:26starseekercomputer aided design
14:56:28kaenI don't know what the BRL part is
14:56:31starseekerhttp://brlcad.org
14:56:52kaenoh cool!
14:56:55kaenwe do gci too :)
14:57:14starseekerwe're using poly2tri for NURBS surface shading right now (generating triangles from the parametric surfaces) but we don't have clipper hooked into the loop for clean-up
14:57:51starseekerlooking into what was available in that department led me to raptor's work with clipper+poly2tri
14:58:07kaenvery cool
14:59:51starseekerif raptor is willing to put together a small github public domain lib that encapsulated the clipper+poly2tri processing subsystem, it should (fingers crossed) slot in where we currently have poly2tri and avoid the nasty fun of crashing the app due to inputs poly2tri can't handle
15:02:51raptorwe had to do many complex things like handle holes in a poly-tree structure
15:03:16starseekerand such a lib would also be a logical focal point for anyone wanting to improve poly2tri robustness as well - the poly2tri devs don't want to add too much complexity because they like the idea of having implementations in many languages: http://code.google.com/p/poly2tri/issues/detail?id=74
15:03:22raptorso our code may be too complex for many cases...
15:03:32raptorwell actually
15:03:38raptorabout poly2tri robustness
15:03:46raptori had some long discussion with one of the devs
15:04:26raptorand some of the robustness problems are due to the nature of the algorithm and cannot be changed without severely hampering the speed of the algorithm
15:04:41starseekerraptor: understood
15:05:07raptoryeah, it was actually startling to find out that the algorithm itself required certain things to be a certain way
15:05:21raptorbecause, believe me, we went through a lot of robustness problems with them...
15:05:50starseekeris that part of the intermediate clean-up step between clipper and poly2tri?
15:05:54raptorI think kaen's and mine stubborness was really got us to where we are today
15:05:59kaenheh
15:06:09raptorstarseeker: yes
15:06:33kaenand I guess they don't want to maintain a cdt->TriangulateSafely() method, since that would be 2*(numberoflanguages) implementations to maintain
15:06:56raptorwe have some crazy cases in our game and I had to add some point-tweaking to get them to work
15:07:29starseekerraptor: I suspect some of the NURBS surfaces out there are likely to generate crazy stuff too - real world data can do that
15:07:43starseekerare your tweaks specific to your individual data sets?
15:08:45Nothing_Much Quit (Quit: Konversation terminated!)
15:09:05Nothing_Much has joined
15:09:17raptornot really, but specific to polygons for botzones
15:09:31starseekerhmm
15:10:14starseeker wonders if it might prove to be the case that botzone restrictions and NURBS trimming loop restrictions have some similarities
15:11:31raptornot sure what NURBS is
15:11:44starseekerNon-Uniform Rational B-Splines
15:11:49starseekerone sec...
15:12:11starseekerhttp://en.wikipedia.org/wiki/Non-uniform_rational_B-spline
15:13:04kaenso I guess you rasterize the splines (with a grid)? and send the polys to p2t?
15:13:21raptori' not sure our stuff can do 3d...
15:13:37starseekerraptor: the parametric domain (where the trimming happens) is 2D
15:14:41starseekertrimming loops are defined in the 2D space to denote "void" areas. When a point in 2D is evaluated to get the final 3D point, we check to see if the 2D point is inside a trimming loop
15:15:09starseekerso for poly2tri, we take the loops (which are usually parametric curves) and generate polylines from them
15:15:29kaenoh, ok
15:15:36raptorthis is some computational geometry I haven't dealt with before
15:15:45raptorbut if it's in 2d, then great!
15:15:50kaenand you end up with arbitrary 2D polylines?
15:15:56raptorwe know 2d over here at #bitfighter :)
15:16:00starseeker:-)
15:16:13kaenthat's pretty much exaclyt what we have to deal with as well
15:16:19starseeker nods
15:16:43starseekerwhen I saw the description of bitfighter's problem domain, it sounded almost exactly like the kind of data we deal with in 2D
15:18:08starseekerwe already use bare poly2tri and get fairly nice shaded displays, but we experience the same weaknesses as well with the poly2tri algorithm's limitations
15:18:59starseekerand those fun asserts in the error case code mean if we *do* hit a bad case, the application goes down with a thud
15:19:33starseeker is trying to teach poly2tri to return an error code or NULL rather than fail, but that touches a lot of the code
15:19:52kaenthere are comments around those that use exceptions instead of asserts
15:19:57kaenso you could catch them
15:19:58kaeniirc
15:20:06raptoryeah... my advice is to not touch poly2tri
15:20:17kaenalso I think assert.h has a define you can use to make it raise instead of abort
15:20:20raptorjust guarantee good input
15:20:34kaen^ we had much better luck with this
15:20:46raptorwe played with poly2tri a little once, got it to not crash... then proceeded to get weird output
15:21:21starseeker nods - I can live with weird output instead of not crashing, but I suppose properly handling the asserts would be better - then we could fall back on a wireframe view
15:21:29kaenah, that's right
15:21:35raptori have to go for a few hours... back later!
15:21:42starseekerraptor: cool - thanks!
15:21:46raptor Quit ()
15:22:00kaenI've done a lot of work on the triangulation as well
15:22:08kaenif you have anymore questions, fire away
15:22:16starseekerkaen: so you see my interest in the possibility of a shared library that encapulates what you have achieved :-)
15:22:39kaenyes indeed
15:23:03starseekerkaen: I'm hoping that I can do exactly that - use pre-processers to guarantee good input (as poly2tri defines good) and avoid ever getting to the abort cases in the first place
15:23:06starseekermuch better all around
15:23:18kaenI guess the only hesitation is that we've only addressed the problems in our specific domain, and have also made assumptions that are probably only valid for us
15:23:39kaennot to mention that our transformations are tied to our own data types (our point class, for example)
15:23:50starseekerkaen: so far, your domain sounds pretty similar
15:23:59starseekeris your point class very complex?
15:24:18kaenno, very simple in fact
15:24:35kaenbut maybe a little weird when it comes to method names
15:25:21starseekerkaen: I think that's probably a surmountable problem then - after all, clipper and poly2tri also have their own point classes
15:25:57starseekercouldn't you just define a separate point class (if need be) that provides the methods needed?
15:26:07kaentrue
15:26:26kaenprobably better would be to put them into either p2t or clipper classes, and do our custom cleanups in those datatypes
15:26:40starseeker nods
15:26:43kaenp2t is probably the better choice, clipper datatypes are a bit... strange
15:26:51starseekerheh - agreed
15:27:17starseekerwhat assumptions do you see as being problematic?
15:28:10kaenhmm I'll have to take a look to refresh my memory. I worry more about the implicit assumptions we didn't realize we were making
15:28:25kaenwe didn't do any sort of formal analysis, of course :D
15:28:41kaenjust lots of punk rock and coffee, for me
15:28:46kaenand a few sundays
15:28:47starseeker<snort> I doubt any such code has that behind it, unless that explains how clipper is so bulletproof...
15:29:01starseekerhey - it's the reciepe that created Linux :-)
15:29:02kaenhaha
15:29:43starseekerkaen: can you think of any such assupmtions you have made that a) you want to keep making rather than address and b) aren't safe assumptions for similar use cases?
15:30:39starseeker got the impression based on what he saw in the various discussions and bug reports that your inputs are actually quite general within this category of geometric problem
15:30:53Nothing_Much Quit (Quit: Konversation terminated!)
15:31:10Nothing_Much has joined
15:32:29kaenlooking over it right now. I'm at work too, it will be a few minutes
15:32:30kaensorry
15:32:41starseekerkaen: no problem
15:34:29starseekerkaen: I should stress that I don't want to push y'all towards doing something you aren't interested in doing
15:42:18kaenwow, funny how c++ looks like chicken scratch after six months of writing ruby
15:42:32kaenanyway, I guess we aren't really making any explicit assumptions
15:42:42kaenother than the inputs not crashing clipper, which is pretty safe
15:43:04starseekerheh
15:43:34kaenother than that it's just simple data transforms between bf, clipper, and p2t with some scaling mixed in
15:43:52starseekerplus the input validity between clipper and p2t?
15:44:01starseekers/input validity/input checking
15:44:07kaenright
15:45:09starseekerso in principle then, a bf->p2t container translation, and a p2t->clipper translation, would allow all the point logic to live in p2t space?
15:45:32starseeker(presumably a p2t->bt conversion already exists...)
15:52:06kaenhmm
15:52:25kaenI think we could templatize it to operate on anything with an x() and y() method
15:52:45kaenoh, and size() and [] for the polygons
15:52:46starseekerah. that would be very general
15:53:07kaenor maybe just use an iterator since we're using stl vectors now
15:53:25kaenthat would save the user at least two manual data transformations
15:53:32starseekerspeed++
15:54:11kaenI think speed-- but usability++
15:54:21kaenso, it actually seems pretty straightforward
15:54:37starseekerhmm - wouldn't avoiding data trasformations be a good thing for speed?
15:55:00kaenit would still be doing the transforms, just that it would save the user from coding it themself
15:55:11starseekeroh, gotcha
15:55:21kaensince it would be datatype-agnostic as long as those methods exist on the source/target
15:55:37starseekeryeah, if they're going to have to the conversions anyway, then it's a definite usability win
15:55:54kaenhmm, I need to turn this over in my head a bit
15:56:11kaenbut I was wrong about the assumptions at least, so I think it makes sense to break this out a bit
15:56:26kaenif it saves even one coder from data transform hell, it's worth it
15:56:31starseeker:-)
15:56:40starseekermy least favorite part of mesh/triangle programming
15:57:26starseekerno, scratch that - second least favorite
15:57:27Nothing_Much Quit (Quit: Konversation terminated!)
15:57:57starseekerleast favorite is trying to see which of the 10,000 possible issues is at play when the OpenGL window isn't showing what I told it to show.
15:58:05kaenoh god
15:58:08kaenI know that pain
15:58:18starseekerbut data transformation is definitely up there
15:58:58starseeker"so, what was your accomplishment today?" uh... discovering 142 ways to create a blank OpenGL view?
15:59:05kaenhahaha
16:00:05starseekerkaen: gotta run myself, be back later this evening most likely. Thank you (and raptor) for taking the time to consider this
16:00:16kaenno problem, see you around
16:00:19kaenthanks for dropping by :)
16:04:55Nothing_Much has joined
16:31:23Nothing_Much Quit (Quit: Konversation terminated!)
16:31:43Nothing_Much has joined
17:30:27Nothing_MuchCan somebody test out my server? :)
17:30:35Nothing_MuchShould be the 24/7 BM and CTF Server
17:37:10Nothing_MuchHey
17:37:17Nothing_MuchI have a Bitfighter .deb
17:37:19Nothing_MuchAnybody need it?
17:45:37Nothing_MuchEverybody's gone now! D:
18:24:04kaen Quit (Read error: Connection reset by peer)
19:15:27Akien Quit (Remote host closed the connection)
19:25:45Nothing_Much Quit (Ping timeout: 240 seconds)
19:29:37Nothing_Much has joined
21:49:34raptor has joined
21:49:34ChanServ sets mode +o
21:49:42raptorgood evening
21:52:10YoshiSmb_m has joined
21:53:07YoshiSmb_mhi guys
22:03:58Nothing_Muchevenin'
22:04:04Nothing_Muchgot my server up 'n runnin'!
22:11:25YoshiSmb_mand, nickserv just fucked up my registered nick
22:33:18starseeker yawns
22:33:59starseekerbusy day
22:34:18YoshiSmb_myea
22:35:02YoshiSmb_msame here, i just got to my new house
22:35:44YoshiSmb_m(and took me like 7 - 8 hours to get everything here)
22:36:04starseekerhmm - "A great event hosted by sky_lark" - is it coincidence or does an E.E. Doc Smith fan lurk among the bitfighters?
22:36:29starseekerYoshiSmb_m: my sympathies - I've had to move several times, and none of them were fun
22:36:49starseekerNothing_Much: I see your server
22:36:59YoshiSmb_msame here
22:37:05YoshiSmb_mim tired
22:37:21YoshiSmb_mbut i will not stop! :-D
22:40:16raptorI'm working on posting the clipper -> poly2tri code
22:41:09starseekerraptor: cool! kaen took a look at what was involved after you left earlier
22:42:51starseekerah, I see your irc logs are pretty much immediately updated - nevermind :-)
22:43:27starseeker(that's nifty, btw - used to seeing logs appear only the next day for some reason on other projects)
22:44:51raptoryeah... i spent a lot of time on that :) but I stole much of the ideas/code from the naev project
23:55:40YoshiSmb_m Quit (Remote host closed the connection)

Index Search ←Prev date Next date→

These logs were automatically created by BFLogBot on irc.freenode.net.