Chat log from the meeting on 2007-10-09

From OpenSimulator

(Difference between revisions)
Jump to: navigation, search
m (Robot: Cosmetic changes)
m
 
Line 1: Line 1:
 
__NOTOC__
 
__NOTOC__
 
{{Quicklinks}}
 
{{Quicklinks}}
<br />
+
<pre style="white-space: pre-wrap;
 +
white-space: -moz-pre-wrap;
 +
white-space: -pre-wrap;
 +
white-space: -o-pre-wrap;
 +
word-wrap: break-word">
 +
[10:45] Stefan Andersson: By the way; if anybody wants to explore the code, writing a simple 'module'
 +
[10:45] Stefan Andersson: that logs all chat in a region to file would be a fun little project
 +
[10:45] Dalien1 Talbot: indeed
 +
[10:45] Tleiades Hax: and usefull for the meetings as well
 +
[10:46] Stefan Andersson: MW introduced the module architecture so that people could write things more or less stand-alone.
 +
[10:46] Dalien1 Talbot: shouldn't we better revive the IRC gateway module ?
 +
[10:46] Dalien1 Talbot: that one was quite functional..
 +
[10:46] Stefan Andersson: Personally, I'm surprised that we haven't had more people setting up custom sims and writing modules.
 +
[10:46] Stefan Andersson: Why not have both?
 +
[10:46] Dalien1 Talbot: true.
 +
[10:47] Stefan Andersson: Everybody seems to want to hax core funcs.
 +
[10:47] Tleiades Hax: the infrastructure for dynamically loading modules isn't in place yet
 +
[10:47] Stefan Andersson: Which is fine; but I think the big breakthru comes the day people start pushing code from their own web sites.
 +
[10:48] Stefan Andersson: Tlei, that's why I want people to get cracking at it. ;)
 +
[10:48] Tleiades Hax: I think two things need to change, maybe three
 +
[10:48] Tleiades Hax: we need a way do dynamically register plugins
 +
[10:48] Tleiades Hax: we need to fill in the wiki page on modules
 +
[10:49] Tleiades Hax: and as a possible last thing, but almost unrelated
 +
[10:49] Stefan Andersson: That should be easy enough; a pick-up directory, or a part of the config.
 +
[10:49] Tleiades Hax: I think we should isolate the plugin pattern we use into an interface
 +
[10:49] Tleiades Hax: I think both
 +
[10:49] Stefan Andersson: Well, yes; the plug-ins should have a well-defined facade into the system
 +
[10:50] Stefan Andersson: that should coincide with the script engine facade
 +
[10:50] Tleiades Hax: scan a subdirectory at run time, and look in the config file
 +
[10:50] Stefan Andersson: As I said, simple enough, I think.
 +
[10:50] Stefan Andersson: We already have dir-scanners and config readers..
 +
[10:50] Tleiades Hax: yes, we have all the component parts
 +
[10:51] Tleiades Hax: should be a piece of cake to do
 +
[10:51] Stefan Andersson: Actually, we have so many of them they should be refactored into generic plug-in handlers
 +
[10:52] Tleiades Hax: true
 +
[10:52] Tleiades Hax: which is why I think we should define an IPlugin interface
 +
[10:56] Stefan Andersson: Definitively
 +
[10:56] Stefan Andersson: Actually, if we could get all our plug-ins into the same structure, so that the different plugin managers could just fetch a list of plug-ins implementing a certain type; that'd be like ace++
 +
[10:57] Neas Bade: welcome all
 +
[10:57] Neas Bade: we actually have a pretty reasonable turn out already
 +
[10:57] Dalien1 Talbot: all of the plugins - like ?
 +
[10:58] Stefan Andersson: So that the IPlugin should have something like IPlugin.Register( IRegisterPluginFacade facade )
 +
[10:58] Dalien1 Talbot: ah something like meta-interface then ?
 +
[10:58] Stefan Andersson: and IRegisterPluginFacade would have something like RegisterPlugin<TPlugin>( TPlugin instance )
 +
[10:58] Neas Bade: man, it will be nice when we have chains and can sit :)
 +
[10:59] Stefan Andersson: That would mean 'Register me as handling calls to interface TPlugin'
 +
[10:59] Neas Bade: will someone volunteer to upload transcript of the office hours?
 +
[11:00] Stefan Andersson: Then any Plugin could ask IPluginFacade for GetHandler<TPlugin>()
 +
[11:00] Tleiades Hax: if I don't crash, I can do it
 +
[11:00] Charles Krinkeb: I'll be the backup.
 +
[11:00] Dalien1 Talbot: I'll also bounce it to the blog as well.
 +
[11:00] Neas Bade: Tleiades, you have logging turned on?
 +
[11:00] Stefan Andersson: Hm. GetHandler<TPlugin>( string name ) I guess
 +
[11:00] Neas Bade: ok, cool
 +
[11:00] Charles Krinkeb: is neas==sean?
 +
[11:00] Neas Bade: yes
 +
[11:00] Charles Krinkeb: got it
 +
[11:00] Neas Bade is Sean Dague
 +
[11:01] Neas Bade: here and in SL
 +
[11:01] Michael Wright: seems like this is going to be everyone
 +
[11:01] Neas Bade: sorry for showing up a little late, dalien / stefan, what did we I wonder into?
 +
[11:01] Stefan Andersson: Um, anybody up for implementing what I just said? :P
 +
[11:01] Neas Bade: yes, I agree
 +
[11:02] Tleiades Hax: hmm. where do I turn on logging?
 +
[11:02] Stefan Andersson: We were just discussing refining the plug-in handling.
 +
[11:02] Stefan Andersson: Ok, so, I'm off for a while....
 +
[11:02] Neas Bade: stefan: ok
 +
[11:02] Tleiades Hax: k
 +
[11:02] Charles Krinkeb: As the host, welcome to Sang. Which I believe to the most stable sim on osgrid. It is checked for functionality every morning, runs on the osgrid server and has not crashed once.
 +
[11:03] Neas Bade: tleides it is under preferences somewhere
 +
[11:04] Michael Wright: really need to fix up the time progressing sometime (night comes a bit too suddenly)
 +
[11:04] Tleiades Hax: think of as being close to equator
 +
[11:04] Neas Bade: Tleiades posted some suggested agenda on the wiki, which seems like as good a place to start as any. Though prior to that, are there other specific things people would like to talk about?
 +
[11:04] Neas Bade: MW: heh :)
 +
[11:05] Neas Bade: one of the things I'd like to throw out there is starting to build a master todo list on the wiki, as I think that will help letting people know where they can contribute bits
 +
[11:06] Tleiades Hax: yes, agreed
 +
[11:06] Neas Bade: man, everyone's quiet :)
 +
[11:06] Dalien1 Talbot: i have crashed a couple of times :)
 +
[11:06] Tleiades Hax: I think we should clean up the 0.5 list
 +
[11:06] Tleiades Hax: make a free for all list of features they'd like
 +
[11:06] Neas Bade: tleiades, I'd like to hear your thoughts on better approaches to the database layer as well
 +
[11:07] Tleiades Hax: and a list of features we plan for 0.5
 +
[11:07] Neas Bade: MW: no worries
 +
[11:07] Tleiades Hax: I like the list the libsl guys have
 +
[11:08] Neas Bade: url?
 +
[11:08] Tleiades Hax: http://www.libsecondlife.org/wiki/Roadmap
 +
[11:09] Tleiades Hax: as for database layer, I think we should consider things in stages
 +
[11:09] Neas Bade: Ok, I just started a Roadmap page on the wiki
 +
[11:09] Tleiades Hax: right now, focus should be on getting stuf to work
 +
[11:09] Neas Bade: I'll take notes into it in "Uncategorized" for now and refine from there
 +
[11:09] Tleiades Hax: kind of brute force
 +
[11:10] Tleiades Hax: inventory is ok for mysql and sdague has sqlite well under control
 +
[11:11] Michael Wright: the asset system is one area that so needs doing, I have been going to do it for weeks now, but as some of you know, for the last few weeks I've been very busy and just haven't had time
 +
[11:11] Tleiades Hax: I have written some unit tests for that, and verified both sqlite and mysql
 +
[11:11] Charles Krinkeb: Is it a correct assumption that inventory database stuff will not affect the sql users and regions table in grid mode?
 +
[11:11] Neas Bade: tleides, well, I think you correctly pointed out that some of the use of DataSets ends up being problematic for larger data (as we load it all into memory)
 +
[11:12] Tleiades Hax: I think mysql will be used for the larger grids, and sqlite for the standalone ones
 +
[11:12] Neas Bade: I'm assuming you are going to take a slightly different approach on the mysql stuff for that?
 +
[11:12] Tleiades Hax: that is my gut feeling, so no big worries I think
 +
[11:12] Neas Bade: if you come up with a better pattern, I'll adapt the sqlite code to match that
 +
[11:13] Tleiades Hax: yes, not quite as elegant as your code, but it won't flood server memory
 +
[11:13] Neas Bade: yeh, that's fine :)
 +
[11:13] Neas Bade: performance counts for something ;)
 +
[11:13] Neas Bade: we can probably pull some patterns out of it and make it both pretty and performant
 +
[11:14] Neas Bade: I'll admit to being pretty ado.net ignorant when I got started on that code, so learning as I go
 +
[11:14] Tleiades Hax: IInventoryData implementation in MySql, is a sample, on how I plan to approach to MySQL
 +
[11:14] Neas Bade: great
 +
[11:14] Neas Bade: also, those unit tests you have, can you check them into the build tree somewhere sensible?
 +
[11:15] Neas Bade: it would be really great to start getting tests in place for key function
 +
[11:15] Tleiades Hax: well, problem with those are, what about programmers who aren't used to working with unit tests
 +
[11:15] Charles Krinkeb: What sort of tests, neas?
 +
[11:15] Neas Bade: we beat them with sticks until they like it :)
 +
[11:15] danxor danx0r: heh
 +
[11:16] Tleiades Hax: I have written some unitttests, for inventory data
 +
[11:16] danxor danx0r: looks like yoga class
 +
[11:16] Neas Bade: :)
 +
[11:16] Charles Krinkeb: Perhaps tests could go on the wiki with simple step-by-step?
 +
[11:16] nebadonb izumi: heheh yea
 +
[11:16] Tleiades Hax: next will be unit tests for assets
 +
[11:16] Michael Wright: yeah doing unit tests are always good.
 +
[11:16] danxor danx0r: hi sorry I'm late -- technical glitches
 +
[11:17] Neas Bade: danxor and I were actually talking about testing earlier, and I was thinking about starting to create a functional test suite by using libsl to create bots and walk them through the environment. It would help with isolating some of the region crossing bugs I think.
 +
[11:18] Dalien1 Talbot: Neas: indeed. plus, with loadtesting as well...
 +
[11:18] Charles Krinkeb: Hmm. 9 avatars, 8 green dots.
 +
[11:18] Tleiades Hax: especially for the database plugins, it is the only way I can think of, which improves the probability of all the plugins working as expected
 +
[11:18] Tleiades Hax: hey, that is a great idea, automated testing of sim functionality
 +
[11:18] Tleiades Hax: I love that concept :-)
 +
[11:18] danxor danx0r: unit tests, ftw
 +
[11:18] Dalien1 Talbot: charles: self does not create the green dot :)
 +
[11:18] Neas Bade: tleiades: agreed
 +
[11:18] danxor danx0r: so would this be something we could all use on a single machine in stand-alone?
 +
[11:18] nebadonb izumi: you dont see yourself charles
 +
[11:19] Neas Bade: danxor, yes, I think so
 +
[11:19] danxor danx0r: minimap is too small to see individual dots
 +
[11:19] danxor danx0r: ahh zoom. nv
 +
[11:19] Tleiades Hax: danxor.. zoom works
 +
[11:19] danxor danx0r: mind
 +
[11:19] Neas Bade: at least that would be the goal
 +
[11:20] danxor danx0r: it's great to finally see what you all look like :-)
 +
[11:20] Tleiades Hax: lol
 +
[11:20] Michael Wright: I think charles is right about the map, I see my yellow dot for myself then 7 green dots (but there is 8 other people here) one person isn't getting a dot
 +
[11:20] Dalien1 Talbot: s/like/alike/ ? :)
 +
[11:20] Neas Bade: heh :)
 +
[11:20] danxor danx0r: sort of like that old "Addicted To Love" video
 +
[11:20] Charles Krinkeb: It builds a sense of community, and that is very good.
 +
[11:20] danxor danx0r: I see yello + 8
 +
[11:20] Dalien1 Talbot: hmm indeed me too.
 +
[11:21] Tleiades Hax: so, should I check the unit tests in?
 +
[11:21] Neas Bade: yes
 +
[11:21] Tleiades Hax: roger that
 +
[11:21] Neas Bade: how do you execute them at the moment?
 +
[11:21] Neas Bade: is it a manual run, or nunit?
 +
[11:22] Tleiades Hax: well, I set my unit test module as executable, and run "nunit-gui"
 +
[11:23] Tleiades Hax: we probably should set up a propper nunit test project
 +
[11:23] Neas Bade: lets start with what you have and work towards that goal
 +
[11:23] Tleiades Hax: ok
 +
[11:24] Tleiades Hax: having my tests in svn will actually help me, I have lost 'em twice already
 +
[11:24] Tleiades Hax: deleted the project folder, to get a fresh copy of opensim, and forgot to backup the tests
 +
[11:24] Dalien1 Talbot: given the speed with which the things change, i think it is good to push the stuff in as early as possible (assuming it does not break the build or functionality)
 +
[11:25] Dalien1 Talbot: even if it is piece by piece
 +
[11:25] Tleiades Hax: yes, small pieces at a time
 +
[11:25] Tleiades Hax: but I think some branches in svn would be a good thing
 +
[11:25] Dalien1 Talbot: hmm then you'd need to maintain more than one
 +
[11:25] Neas Bade: svn branches are actually a beast to merge
 +
[11:26] Michael Wright: we used to have branches as generally it just leads to problems
 +
[11:26] Dalien1 Talbot: i would rather try to keep it contained within one branch
 +
[11:26] Michael Wright: as= and
 +
[11:26] Neas Bade: I was thinking of creating a mercurial clone of svn trunk for people that wanted to do local branch changes
 +
[11:26] Charles Krinkeb: We declare a moratorium on updates from time to time for a few revisions if we need to pursue an idea that breaks the build for the general user.
 +
[11:27] Dalien1 Talbot: Charles: my build system always keeps the latest successful build
 +
[11:27] danxor danx0r: speaking of svn, I was thinking of creating opensim-libs
 +
[11:27] Neas Bade: in my experience using a distributed scm like mercurial makes life much easier if you want to do lots of branch stuff
 +
[11:27] Dalien1 Talbot: but ideally the build should never break
 +
[11:27] danxor danx0r: and starting down the path of getting a proper 3rd party build
 +
[11:27] Michael Wright: yeah, having a stable tag branch , that is only updated say once a week or whatever, but we tell everyone (non developers) to stick to that, is fine
 +
[11:27] danxor danx0r: what's mercurial? sounds like git
 +
[11:28] Neas Bade: danxor, good point
 +
[11:28] Charles Krinkeb: I was responding to Dan's comment a day or so ago that demanding the build always works may put the developers in a bind from time to time.
 +
[11:28] Neas Bade: danxor, yes, like git, but easier to learn to use
 +
[11:28] Dalien1 Talbot: MW: so that branch would be always re-synced automagically ? or needs manual maintenance ?
 +
[11:28] danxor danx0r: it'll need a merge every week sounds like
 +
[11:28] danxor danx0r: or maybe not
 +
[11:28] danxor danx0r: just replace it with the latest thing
 +
[11:28] Dalien1 Talbot: hmm that would mean manual maintenance... = pita.
 +
[11:29] danxor danx0r: hmmm
 +
[11:29] Dalien1 Talbot: could do simpler imho
 +
[11:29] Neas Bade: is this discussion on the stable tag, or on mercurial?
 +
[11:29] Dalien1 Talbot: just declare that the stuff should work on monday morning build or such
 +
[11:29] Tleiades Hax: stable tag, should be pretty easy to do, I think
 +
[11:29] danxor danx0r: if testing really works, you could build nightly, test, and update stable if test passes
 +
[11:30] Neas Bade: we need a few tests before we could believe that :)
 +
[11:30] danxor danx0r: good way to exercise the unit tests; if they miss something, we'll hear right away from users
 +
[11:30] Neas Bade: but that does tie nicely into test automation to move the stable branch
 +
[11:30] Dalien1 Talbot: danx0r: that's almost what i do on opensim.be, except i do not have the unittests - merely if the compile succeeds then i update with the new binary...
 +
[11:31] Neas Bade: http://opensimulator.org/wiki/Roadmap
 +
[11:31] Charles Krinkeb: Dalien, best to test that the client logs into the new binary to be sure.
 +
[11:31] Neas Bade: been trying to update that on the fly with things poping into my head
 +
[11:32] Dalien1 Talbot: ckrinke: indeed. once there are automated tests for that, I will include them.
 +
[11:32] Tleiades Hax: those automated tests, using libsl would be soooo cool
 +
[11:32] Dalien1 Talbot: although this boils down a bit to the unit testing
 +
[11:32] Dalien1 Talbot: i assume that every person that commits the code does test that the code actually works :)
 +
[11:32] Neas Bade: yes, I think automated tests with libsl will help a lot. I'm going to start digging into that tomorrow
 +
[11:33] Dalien1 Talbot: there were a couple of glitches - but they are usually caught by the build failures, it seems.
 +
[11:33] Tleiades Hax: the problem is testing on all platforms
 +
[11:33] Charles Krinkeb: I believe it is fair to say that code committed denotes responsibility to ensure users on both Linux & Windows have not gone backwards.
 +
[11:33] Dalien1 Talbot: ideally we should have an autobuild for both linux + windows
 +
[11:34] Neas Bade: right, well we need some automation there first, then we can see about that
 +
[11:34] Dalien1 Talbot: so that the windows folks did not have to bother with the linux setup, and vice versa.
 +
[11:34] Neas Bade: we've got opensimulator.org for the linux side
 +
[11:34] chill ken: is the stable tag idea due to time between releases being too long? what about doing more minor releases?
 +
[11:34] Dalien1 Talbot: neas: ah you started to do the builds there as well ?
 +
[11:34] Neas Bade: no, I haven't
 +
[11:34] Neas Bade: but we could
 +
[11:34] danxor danx0r: charles: not sure it's reasonable to expect every commit to include full windows + linux tests
 +
[11:34] danxor danx0r: unless they are automated
 +
[11:34] Dalien1 Talbot: ah ok - because http://ruth.opensim.be/build does do nightly build (nightly CET :)
 +
[11:35] danxor danx0r: I try to use my intuition about how far to test each commit
 +
[11:35] Neas Bade: yeh, and I think that's fair
 +
[11:35] Dalien1 Talbot: yeah i think it is a good approach.
 +
[11:35] Neas Bade: if people are running off svn they need to know there are risks there
 +
[11:35] Charles Krinkeb: No, but I would suggest it is reasonable that each commit, makes a commitment to ensure the users have moved forward and not backwards. That can be as simple as making sure at least one user compiles and runs on the other platform.
 +
[11:35] Tleiades Hax: I think almost all, run off svn head
 +
[11:35] danxor danx0r: that's exactly why there should be a semi-stable branch for the increasing number of 'power users' who compile but don't typically code
 +
[11:36] Dalien1 Talbot: Charles: I do not have the windows environment, and some folks do not have linux.
 +
[11:36] Neas Bade: well, I'm actually leaning against the stable tag, as it ends up being a crutch for not releasing more often
 +
[11:36] Charles Krinkeb: I would concur with Sean
 +
[11:36] Dalien1 Talbot: danx0r: maintaining more than one branch is a pain..
 +
[11:36] danxor danx0r: yeah, I think the defacto assumption has been, test n your own machine
 +
[11:36] Tleiades Hax: I'm leaning towards that point of view too
 +
[11:37] danxor danx0r: typically if you're not doing something involving build or native libs, you won't break the other platform
 +
[11:37] danxor danx0r: though I did have an exception where mono didn't run something .NET did
 +
[11:37] Tleiades Hax: the biggest pain is in the native libs
 +
[11:37] danxor danx0r: I suppose we could run mono on windows, and say mono is the gold standard for "this build doesn't segfault"
 +
[11:37] Neas Bade: I think that people are running on svn because the code is moving fast. We can't keep the code moving fast if we require a full regression test on every checkin :)
 +
[11:37] Dalien1 Talbot: Neas: indeed
 +
[11:38] danxor danx0r: Tleiades -- I want to talk about the lib problem
 +
[11:38] Neas Bade: yeh, native libs is a beast
 +
[11:38] Neas Bade: danxor: go for it
 +
[11:38] danxor danx0r: I think *all* libs, native or managed, that are not par tof standard build should be in a separate project
 +
[11:38] danxor danx0r: it will have two subfolders, managed & unmanaged
 +
[11:38] danxor danx0r: and each one has folders for each lib
 +
[11:39] danxor danx0r: then a top-level script to (eventually) invoke all the makes
 +
[11:39] danxor danx0r: (though some are VS hand-make only now)
 +
[11:39] Neas Bade: I like that approach
 +
[11:39] danxor danx0r: copy the dll's & so's into bin
 +
[11:39] danxor danx0r: and perhaps create zips -- tho that gets into architecture
 +
[11:39] Neas Bade: I think it would make sense not to polute bin/ with them though
 +
[11:39] Neas Bade: it would be nice to have a seperate lib/
 +
[11:39] danxor danx0r: ie, someone somewhere needs to build them for each target supported
 +
[11:40] danxor danx0r: neas (sdague?) -- dunno how mono .net deals with looking for them
 +
[11:40] danxor danx0r: but keep in mind stuff like ode.net.dll needs to find the unmanged ode.dll
 +
[11:40] danxor danx0r: AND, sometimes ode.dll will need to find other native stuff
 +
[11:40] Tleiades Hax: mono is actually easier than .net I think
 +
[11:40] Neas Bade: you just need to add paths in the prebuild.xml
 +
[11:40] Neas Bade: or nant
 +
[11:40] danxor danx0r: well I don't mind polluting bin, if it's not in subversion
 +
[11:41] danxor danx0r: ok I didn't explain right --
 +
[11:41] danxor danx0r: you can choose to build libs, _OR_ you just donwload a zip file
 +
[11:41] danxor danx0r: which someone built for your platform
 +
[11:41] danxor danx0r: and that zips into bin, or ./lib if you prefer
 +
[11:41] danxor danx0r: no binaries in subversion!
 +
[11:41] Neas Bade: well, what you are really talking about is opensim-libs releases
 +
[11:41] danxor danx0r: yes, right
 +
[11:41] Neas Bade: yep
 +
[11:42] Neas Bade: I'm all for that
 +
[11:42] danxor danx0r: but it's not going to be on the same schedule as opensim commits
 +
[11:42] Tleiades Hax: I like that too
 +
[11:42] Neas Bade: opensim-libs is an svn repo
 +
[11:42] danxor danx0r: yup
 +
[11:42] danxor danx0r: we can start with just reasonable file organization
 +
[11:42] Neas Bade: we can make it an svn external to the main opensim so you can have a single checkout to get the all if you are buliding from source
 +
[11:42] danxor danx0r: and the big makefile, all hail the make can come later
 +
[11:42] Neas Bade: well, probably we want nant instead of make
 +
[11:42] danxor danx0r: because for now, you gotsta do all the README's and INSTALL.txt
 +
[11:42] danxor danx0r: for each liv
 +
[11:43] danxor danx0r: lib
 +
[11:43] danxor danx0r: but at least, these will be the proper versions,
 +
[11:43] danxor danx0r: with maybe an opensim README about compile opts
 +
[11:43] Neas Bade: ok, danxor, you got the ball for that one?
 +
[11:43] danxor danx0r: so you can walk through it and end up at aknown state
 +
[11:43] Dalien1 Talbot: hm how do we "freeze" the version of the foreign lib ?
 +
[11:43] danxor danx0r: yeah (sigh)
 +
[11:43] Neas Bade: to do the new repo
 +
[11:43] danxor danx0r: I'll start it
 +
[11:43] Dalien1 Talbot: e.g. libsl
 +
[11:44] danxor danx0r: what else besides ode do I need to build?
 +
[11:44] danxor danx0r: libsl,
 +
[11:44] Neas Bade: libsl
 +
[11:44] danxor danx0r: is taht C++?
 +
[11:44] Dalien1 Talbot: if we just have it "pull the latest"
 +
[11:44] Neas Bade: it's libopenjpeg is the .so
 +
[11:44] danxor danx0r: did we get the version thing right now?
 +
[11:44] danxor danx0r: dalien -- "pull the latest" is a problem --
 +
[11:44] Dalien1 Talbot: then the changes in the libsl would break the opensim-libs
 +
[11:44] Dalien1 Talbot: precisely
 +
[11:44] Neas Bade: dalien, you can set a version number on an svn external IIRC
 +
[11:44] danxor danx0r: that's why we need copies of everything -- so we can say
 +
[11:44] danxor danx0r: we know this works with opensim
 +
[11:44] Neas Bade: sort of like a tag
 +
[11:44] danxor danx0r: you want to pull unstable and take a bet, go ahead
 +
[11:44] Dalien1 Talbot: or we svn co -r
 +
[11:45] Dalien1 Talbot: particular revision number
 +
[11:45] Neas Bade: http://svnbook.red-bean.com/en/1.0/ch07s03.html
 +
[11:45] Tleiades Hax: we are caught inbetween chairs on libsl
 +
[11:45] danxor danx0r: dalien -- well whoever maintains it needs to document what they do
 +
[11:45] danxor danx0r: there may be patches or subtle make changes
 +
[11:45] Dalien1 Talbot: cos we'd ideally then need the universal approach to all of the libs
 +
[11:45] danxor danx0r: for instance, I modified ODE's build to statically link those stupid msvc dll's
 +
[11:46] danxor danx0r: but at least it's there in svn, in a known state
 +
[11:46] Dalien1 Talbot: right... so could do something like gentoo & co - download the version that is needed, unpack, possibly patch, build ?
 +
[11:46] Tleiades Hax: msvc dll dependencies are always a pain
 +
[11:46] danxor danx0r: if someone does somethhing funky, yeah you just svn -r opensim-libs
 +
[11:46] danxor danx0r: nice
 +
[11:46] Stefan Andersson: Ok, so back.
 +
[11:46] Neas Bade: yeh, getting there would be great
 +
[11:46] danxor danx0r: tl -- turns out you can just compile statically but no one does
 +
[11:46] Stefan Andersson: Wow, quite the turnout. And no sim crashes.
 +
[11:46] danxor danx0r: ok I take the ball on this, with my copiouis free time
 +
[11:47] Tleiades Hax: yes, we do that at my workplace
 +
[11:47] Stefan Andersson waves YAY!
 +
[11:47] danxor danx0r: I'd love to have another commit-level dev on physics...
 +
[11:47] danxor danx0r: no crash because we all stand still with our arms out
 +
[11:47] Neas Bade: speaking of physics, how is that all going?
 +
[11:47] danxor danx0r: and act like it's IRC
 +
[11:47] danxor danx0r: physics update: got a major bug with terrain
 +
[11:47] danxor danx0r: strange one, hunting it down
 +
[11:47] Charles Krinkeb: Yep, you'all made a very stable sim server.
 +
[11:48] danxor danx0r: then some tweaks on bouncinetss
 +
[11:48] danxor danx0r: bounciness
 +
[11:48] paulie Femto: Howdy hi!
 +
[11:48] Neas Bade: hi paulie
 +
[11:48] danxor danx0r: darok made some bullet improvements but it needs real work
 +
[11:48] Tleiades Hax: hi paulie
 +
[11:48] danxor danx0r: Gerard did that awesome hollow patch, but it's one of 20+ prim types
 +
[11:48] danxor danx0r: all of which can be in combination -- so that's a ton of work
 +
[11:48] Neas Bade: are there specific things others can do to help out?
 +
[11:49] danxor danx0r: I'd like to pull GErard in if possible
 +
[11:49] danxor danx0r: well...
 +
[11:50] Tleiades Hax: at some point, work on assets will overlap with the prim types
 +
[11:50] Neas Bade: right, that's a good point tleiades
 +
[11:50] danxor danx0r: help: prim types
 +
[11:50] danxor danx0r: yes
 +
[11:50] danxor danx0r: it's some wicked stuff
 +
[11:51] danxor danx0r: common code to manage it would be great. Not sure who needs to know other than physics + client tho
 +
[11:51] danxor danx0r: scripts?
 +
[11:51] Tleiades Hax: scripts will need it
 +
[11:51] Tleiades Hax: to change prims based on scripts
 +
[11:51] Tleiades Hax: and those changes needs to be persisted into the asset store
 +
[11:52] Tleiades Hax: I scanned the ll functions today, and all of those functions are not implemented
 +
[11:52] Dalien1 Talbot: MW/lbsa: would like to spend 3-4 minutes with you later to discuss the TP stuff...(not to interrupt the physics thread now)
 +
[11:53] Charles Krinkeb: TL. Only about a third are implemented.
 +
[11:53] Neas Bade: right, I think actually getting assets somewhat more functional is probably required first
 +
[11:53] Tleiades Hax: yes, I noticed, and for pretty good reasons too
 +
[11:53] Neas Bade: which I'm very excited tleiades is digging into so much :)
 +
[11:53] Tleiades Hax: I plan on working on that, for a little while
 +
[11:53] Michael Wright: Dalien, okay but think for me at least, its going to have to be tomorrow, been a long day and need to spend some time resting
 +
[11:53] danxor danx0r: other physics issues:
 +
[11:54] danxor danx0r: in my opinion, ODE will not cut it without it being forked and worked on
 +
[11:54] danxor danx0r: it's just too damn unstable numerically
 +
[11:54] Dalien1 Talbot: ok.. actually i think i will need to discuss with Tleiades too :-)
 +
[11:54] danxor danx0r: it throws exceptions that need to be handled in C++
 +
[11:54] danxor danx0r: ppl have recommended OGre and other things
 +
[11:54] paulie Femto: Hows Bullet comin along?
 +
[11:54] Neas Bade: once we get assets working to a reasonable amount, there are a few other folks that I work with that are happy to grind out LSL functions
 +
[11:54] danxor danx0r: you sort of need a 'game engine' on top of physics, I've been ad-hacking it
 +
[11:55] danxor danx0r: bullet would need similar investment
 +
[11:55] danxor danx0r: but at least it's C#
 +
[11:55] paulie Femto: BulletX is more managed from the start, yeah.
 +
[11:55] Neas Bade: danxor, can you explain the difference between what we physics and game engine?
 +
[11:55] Neas Bade: in terms of what opensim needs?
 +
[11:56] danxor danx0r: well
 +
[11:56] danxor danx0r: phsics is very general
 +
[11:57] danxor danx0r: objects, vectors, momentum, collision
 +
[11:57] danxor danx0r: game engines add character classes
 +
[11:57] danxor danx0r: terrain
 +
[11:57] danxor danx0r: ability to modify & move stuff around without all hell breaking loose
 +
[11:57] danxor danx0r: the character class is probably the most important
 +
[11:57] danxor danx0r: and vehicle thingies
 +
[11:57] danxor danx0r: it can all be done from scratch, you know, just another layer of code to conceptualize and debug
 +
[11:58] danxor danx0r: I started out thinking it would be piece of cake
 +
[11:58] danxor danx0r: like everything, turns out it's sort of hard & stuff
 +
[11:58] Neas Bade: what do you think the best approach here is in terms of getting there sooner rather than later?
 +
[11:58] Neas Bade: :)
 +
[11:58] danxor danx0r: so if someone did it, I'd rather not have to
 +
[11:58] Neas Bade: what, it's not SMOP?
 +
[11:58] danxor danx0r: I think I need to look into what's available for C#
 +
[11:58] danxor danx0r: rather for .NET I guess
 +
[11:58] danxor danx0r: SMOP?
 +
[11:59] Neas Bade: simple matter of programming
 +
[11:59] danxor danx0r: sorry felt the need to collide
 +
[11:59] Tleiades Hax: no physics here
 +
[11:59] danxor danx0r: heh
 +
[11:59] Charles Krinkeb: basicphysics only on Sang
 +
[11:59] danxor danx0r: oh well
 +
[11:59] danxor danx0r: so can I walk thru you?
 +
[11:59] danxor danx0r: ok
 +
[12:00] danxor danx0r: hah SMOP exactly
 +
[12:00] Neas Bade: hehehe
 +
[12:00] danxor danx0r: frankly before this project, I poo-pooed "game engines"
 +
[12:00] Charles Krinkeb: "mere implementation detail"?
 +
[12:00] Tleiades Hax: can we break the physics thing into smaller pieces?
 +
[12:00] danxor danx0r: I figured I'm a macho programmer, etc
 +
[12:00] paulie Femto: Has anyone seen nixnerd around? Anyon eknow if he solved his issues with PHP launching a screen session?
 +
[12:00] paulie Femto: brb. door.
 +
[12:02] Charles Krinkeb: Sean: Do you envision this as a recurring weekly meeting and if so, where do we meet next Tuesday?
 +
[12:02] Neas Bade: ok, I need to run away for a bit and pick up food at the farm here. Can people look at the Roadmap I started at: http://opensimulator.org/wiki/Roadmap and embelish with more
 +
[12:02] Neas Bade: yes, I think we should try to do this every week
 +
[12:02] danxor danx0r: tleiades: one issue with physics is it's wired into the main code kinda slapdash
 +
[12:03] Neas Bade: given that we've all gotten a bit more scattered on IRC, it is helpful to have once a week of actually people all being roughly active at the same time
 +
[12:03] danxor danx0r: there was talk pre 0.4 of a proper event model
 +
[12:03] Charles Krinkeb: This sim is always up
 +
[12:03] Neas Bade: want to repropose event models on the -dev list?
 +
[12:03] Tleiades Hax listens
 +
[12:03] danxor danx0r: but we'd be more productive chatting if SL chat had name completion like IRC
 +
[12:04] Neas Bade: yeh, maybe :)
 +
[12:04] Dalien1 Talbot: danx0r: it's a client feature on IRC :)
 +
[12:04] Tleiades Hax: meeting here, is a good way of keeping track of our progress
 +
[12:04] Michael Wright: I think the tuesday meetings every week is good, its a little bit late for me, but we aren't going to get times that are good for everyone, and I guess one bonus of it being at this time on a tuesday is that if we stick to a hour, then Zero's office hour is straight after (ie its just starting now)
 +
[12:04] Neas Bade: yeh, that was also a sneaking idea by me :)
 +
[12:04] danxor danx0r: have LL said anything about physics in the WG?
 +
[12:04] danxor danx0r: it seems to get short shrift
 +
[12:04] Neas Bade: nope, not yet
 +
[12:05] danxor danx0r: like "yeah, we'll add some physics spice at the end"
 +
[12:05] Stefan Andersson: Yeah; it's a bit late for me too, I can't do any real work, but as MW said, it's good enough.
 +
[12:05] danxor danx0r: it shoud be a core ingredient, not a condiment :)
 +
[12:05] Charles Krinkeb: The best meetings have a beginning, a middle, and most important, an end.
 +
[12:05] Neas Bade: honestly, the lindens haven't been big on details for the AWG yet
 +
[12:05] Stefan Andersson: Alas, I need to go to bed, as I'm supposed to rise in like 5 hours.
 +
[12:05] Tleiades Hax: I am so totally at odds with the AWG
 +
[12:05] Stefan Andersson: Bye all, ace seeing you all on Sang.
 +
[12:05] Tleiades Hax: nite
 +
[12:05] Neas Bade: nite
 +
[12:05] danxor danx0r: well we need to do some wiki work on 0.5
 +
[12:05] Charles Krinkeb: nite all
 +
[12:05] Dalien1 Talbot: ok cya Stefan
 +
[12:06] danxor danx0r: are we all agreed that there be some sort of physics officially supported?
 +
[12:06] Stefan Andersson: (Sang's proabbly gonna be that legendary 'sirst island' that people talk about in SL)
 +
[12:06] Neas Bade: well, get over to SL now for Zero's office hours, and you can probably bring it up
 +
[12:06] danxor danx0r: ok
 +
[12:06] danxor danx0r: I'll probably just listen for a bit
 +
[12:06] Neas Bade: I took a screen shot
 +
[12:06] danxor danx0r: are we breaking up already?
 +
[12:06] Tleiades Hax: yes
 +
[12:06] Dalien1 Talbot: Neas: I'll ping you on SL for a TP ?
 +
[12:06] Dalien1 Talbot: I will blog the meeting :)
 +
[12:06] Neas Bade: I plan to do it every week, so we can have an animated progression of opensim
 +
[12:06] danxor danx0r: you look Mahvleous!
 +
[12:06] Neas Bade: catch you all later
 +
[12:07] danxor danx0r: later
 +
[12:07] Neas Bade: dalien, sure
 +
[12:07] Tleiades Hax: danxor, I think a lot will want to catch Zero Lindens office hours
 +
[12:07] danxor danx0r: where is zero's office
 +
[12:07] Tleiades Hax: hmm... I forgot
 +
[12:07] Dalien1 Talbot: danx0r: ping me on SL in a few - dalien talbot, once I get there via Sean, I TP you too :)
 +
[12:07] Dalien1 Talbot: cya on SL...
 +
[12:07] danxor danx0r: ok
 +
[12:08] Tleiades Hax: should we reconvene?
 +
[12:08] Dalien1 Talbot: lets continue in IM on the SL ?
 +
[12:08] Dalien1 Talbot: (and could do a group there, i suppose?)
 +
</pre>
  
[10:45] Stefan Andersson: By the way; if anybody wants to explore the code, writing a simple 'module'<br />
 
[10:45] Stefan Andersson: that logs all chat in a region to file would be a fun little project<br />
 
[10:45] Dalien1 Talbot: indeed<br />
 
[10:45] Tleiades Hax: and usefull for the meetings as well<br />
 
[10:46] Stefan Andersson: MW introduced the module architecture so that people could write things more or less stand-alone.<br />
 
[10:46] Dalien1 Talbot: shouldn't we better revive the IRC gateway module ?<br />
 
[10:46] Dalien1 Talbot: that one was quite functional..<br />
 
[10:46] Stefan Andersson: Personally, I'm surprised that we haven't had more people setting up custom sims and writing modules.<br />
 
[10:46] Stefan Andersson: Why not have both?<br />
 
[10:46] Dalien1 Talbot: true.<br />
 
[10:47] Stefan Andersson: Everybody seems to want to hax core funcs.<br />
 
[10:47] Tleiades Hax: the infrastructure for dynamically loading modules isn't in place yet<br />
 
[10:47] Stefan Andersson: Which is fine; but I think the big breakthru comes the day people start pushing code from their own web sites.<br />
 
[10:48] Stefan Andersson: Tlei, that's why I want people to get cracking at it. ;)<br />
 
[10:48] Tleiades Hax: I think two things need to change, maybe three<br />
 
[10:48] Tleiades Hax: we need a way do dynamically register plugins<br />
 
[10:48] Tleiades Hax: we need to fill in the wiki page on modules<br />
 
[10:49] Tleiades Hax: and as a possible last thing, but almost unrelated<br />
 
[10:49] Stefan Andersson: That should be easy enough; a pick-up directory, or a part of the config.<br />
 
[10:49] Tleiades Hax: I think we should isolate the plugin pattern we use into an interface<br />
 
[10:49] Tleiades Hax: I think both<br />
 
[10:49] Stefan Andersson: Well, yes; the plug-ins should have a well-defined facade into the system<br />
 
[10:50] Stefan Andersson: that should coincide with the script engine facade<br />
 
[10:50] Tleiades Hax: scan a subdirectory at run time, and look in the config file<br />
 
[10:50] Stefan Andersson: As I said, simple enough, I think.<br />
 
[10:50] Stefan Andersson: We already have dir-scanners and config readers..<br />
 
[10:50] Tleiades Hax: yes, we have all the component parts<br />
 
[10:51] Tleiades Hax: should be a piece of cake to do<br />
 
[10:51] Stefan Andersson: Actually, we have so many of them they should be refactored into generic plug-in handlers<br />
 
[10:52] Tleiades Hax: true<br />
 
[10:52] Tleiades Hax: which is why I think we should define an IPlugin interface<br />
 
[10:56] Stefan Andersson: Definitively<br />
 
[10:56] Stefan Andersson: Actually, if we could get all our plug-ins into the same structure, so that the different plugin managers could just fetch a list of plug-ins implementing a certain type; that'd be like ace++<br />
 
[10:57] Neas Bade: welcome all<br />
 
[10:57] Neas Bade: we actually have a pretty reasonable turn out already<br />
 
[10:57] Dalien1 Talbot: all of the plugins - like ?<br />
 
[10:58] Stefan Andersson: So that the IPlugin should have something like IPlugin.Register( IRegisterPluginFacade facade )<br />
 
[10:58] Dalien1 Talbot: ah something like meta-interface then ?<br />
 
[10:58] Stefan Andersson: and IRegisterPluginFacade would have something like RegisterPlugin<TPlugin>( TPlugin instance )<br />
 
[10:58] Neas Bade: man, it will be nice when we have chains and can sit :)<br />
 
[10:59] Stefan Andersson: That would mean 'Register me as handling calls to interface TPlugin'<br />
 
[10:59] Neas Bade: will someone volunteer to upload transcript of the office hours?<br />
 
[11:00] Stefan Andersson: Then any Plugin could ask IPluginFacade for GetHandler<TPlugin>()<br />
 
[11:00] Tleiades Hax: if I don't crash, I can do it<br />
 
[11:00] Charles Krinkeb: I'll be the backup.<br />
 
[11:00] Dalien1 Talbot: I'll also bounce it to the blog as well.<br />
 
[11:00] Neas Bade: Tleiades, you have logging turned on?<br />
 
[11:00] Stefan Andersson: Hm. GetHandler<TPlugin>( string name ) I guess<br />
 
[11:00] Neas Bade: ok, cool<br />
 
[11:00] Charles Krinkeb: is neas==sean?<br />
 
[11:00] Neas Bade: yes<br />
 
[11:00] Charles Krinkeb: got it<br />
 
[11:00] Neas Bade is Sean Dague<br />
 
[11:01] Neas Bade: here and in SL<br />
 
[11:01] Michael Wright: seems like this is going to be everyone<br />
 
[11:01] Neas Bade: sorry for showing up a little late, dalien / stefan, what did we I wonder into?<br />
 
[11:01] Stefan Andersson: Um, anybody up for implementing what I just said? :P<br />
 
[11:01] Neas Bade: yes, I agree<br />
 
[11:02] Tleiades Hax: hmm. where do I turn on logging?<br />
 
[11:02] Stefan Andersson: We were just discussing refining the plug-in handling.<br />
 
[11:02] Stefan Andersson: Ok, so, I'm off for a while....<br />
 
[11:02] Neas Bade: stefan: ok<br />
 
[11:02] Tleiades Hax: k<br />
 
[11:02] Charles Krinkeb: As the host, welcome to Sang. Which I believe to the most stable sim on osgrid. It is checked for functionality every morning, runs on the osgrid server and has not crashed once.<br />
 
[11:03] Neas Bade: tleides it is under preferences somewhere<br />
 
[11:04] Michael Wright: really need to fix up the time progressing sometime (night comes a bit too suddenly)<br />
 
[11:04] Tleiades Hax: think of as being close to equator<br />
 
[11:04] Neas Bade: Tleiades posted some suggested agenda on the wiki, which seems like as good a place to start as any. Though prior to that, are there other specific things people would like to talk about?<br />
 
[11:04] Neas Bade: MW: heh :)<br />
 
[11:05] Neas Bade: one of the things I'd like to throw out there is starting to build a master todo list on the wiki, as I think that will help letting people know where they can contribute bits<br />
 
[11:06] Tleiades Hax: yes, agreed<br />
 
[11:06] Neas Bade: man, everyone's quiet :)<br />
 
[11:06] Dalien1 Talbot: i have crashed a couple of times :)<br />
 
[11:06] Tleiades Hax: I think we should clean up the 0.5 list<br />
 
[11:06] Tleiades Hax: make a free for all list of features they'd like<br />
 
[11:06] Neas Bade: tleiades, I'd like to hear your thoughts on better approaches to the database layer as well<br />
 
[11:07] Tleiades Hax: and a list of features we plan for 0.5<br />
 
[11:07] Neas Bade: MW: no worries<br />
 
[11:07] Tleiades Hax: I like the list the libsl guys have<br />
 
[11:08] Neas Bade: url?<br />
 
[11:08] Tleiades Hax: http://www.libsecondlife.org/wiki/Roadmap<br />
 
[11:09] Tleiades Hax: as for database layer, I think we should consider things in stages<br />
 
[11:09] Neas Bade: Ok, I just started a Roadmap page on the wiki<br />
 
[11:09] Tleiades Hax: right now, focus should be on getting stuf to work<br />
 
[11:09] Neas Bade: I'll take notes into it in "Uncategorized" for now and refine from there<br />
 
[11:09] Tleiades Hax: kind of brute force<br />
 
[11:10] Tleiades Hax: inventory is ok for mysql and sdague has sqlite well under control<br />
 
[11:11] Michael Wright: the asset system is one area that so needs doing, I have been going to do it for weeks now, but as some of you know, for the last few weeks I've been very busy and just haven't had time<br />
 
[11:11] Tleiades Hax: I have written some unit tests for that, and verified both sqlite and mysql<br />
 
[11:11] Charles Krinkeb: Is it a correct assumption that inventory database stuff will not affect the sql users and regions table in grid mode?<br />
 
[11:11] Neas Bade: tleides, well, I think you correctly pointed out that some of the use of DataSets ends up being problematic for larger data (as we load it all into memory)<br />
 
[11:12] Tleiades Hax: I think mysql will be used for the larger grids, and sqlite for the standalone ones<br />
 
[11:12] Neas Bade: I'm assuming you are going to take a slightly different approach on the mysql stuff for that?<br />
 
[11:12] Tleiades Hax: that is my gut feeling, so no big worries I think<br />
 
[11:12] Neas Bade: if you come up with a better pattern, I'll adapt the sqlite code to match that<br />
 
[11:13] Tleiades Hax: yes, not quite as elegant as your code, but it won't flood server memory<br />
 
[11:13] Neas Bade: yeh, that's fine :)<br />
 
[11:13] Neas Bade: performance counts for something ;)<br />
 
[11:13] Neas Bade: we can probably pull some patterns out of it and make it both pretty and performant<br />
 
[11:14] Neas Bade: I'll admit to being pretty ado.net ignorant when I got started on that code, so learning as I go<br />
 
[11:14] Tleiades Hax: IInventoryData implementation in MySql, is a sample, on how I plan to approach to MySQL<br />
 
[11:14] Neas Bade: great<br />
 
[11:14] Neas Bade: also, those unit tests you have, can you check them into the build tree somewhere sensible?<br />
 
[11:15] Neas Bade: it would be really great to start getting tests in place for key function<br />
 
[11:15] Tleiades Hax: well, problem with those are, what about programmers who aren't used to working with unit tests<br />
 
[11:15] Charles Krinkeb: What sort of tests, neas?<br />
 
[11:15] Neas Bade: we beat them with sticks until they like it :)<br />
 
[11:15] danxor danx0r: heh<br />
 
[11:16] Tleiades Hax: I have written some unitttests, for inventory data<br />
 
[11:16] danxor danx0r: looks like yoga class<br />
 
[11:16] Neas Bade: :)<br />
 
[11:16] Charles Krinkeb: Perhaps tests could go on the wiki with simple step-by-step?<br />
 
[11:16] nebadonb izumi: heheh yea<br />
 
[11:16] Tleiades Hax: next will be unit tests for assets<br />
 
[11:16] Michael Wright: yeah doing unit tests are always good.<br />
 
[11:16] danxor danx0r: hi sorry I'm late -- technical glitches<br />
 
[11:17] Neas Bade: danxor and I were actually talking about testing earlier, and I was thinking about starting to create a functional test suite by using libsl to create bots and walk them through the environment. It would help with isolating some of the region crossing bugs I think.<br />
 
[11:18] Dalien1 Talbot: Neas: indeed. plus, with loadtesting as well...<br />
 
[11:18] Charles Krinkeb: Hmm. 9 avatars, 8 green dots.<br />
 
[11:18] Tleiades Hax: especially for the database plugins, it is the only way I can think of, which improves the probability of all the plugins working as expected<br />
 
[11:18] Tleiades Hax: hey, that is a great idea, automated testing of sim functionality<br />
 
[11:18] Tleiades Hax: I love that concept :-)<br />
 
[11:18] danxor danx0r: unit tests, ftw<br />
 
[11:18] Dalien1 Talbot: charles: self does not create the green dot :)<br />
 
[11:18] Neas Bade: tleiades: agreed<br />
 
[11:18] danxor danx0r: so would this be something we could all use on a single machine in stand-alone?<br />
 
[11:18] nebadonb izumi: you dont see yourself charles<br />
 
[11:19] Neas Bade: danxor, yes, I think so<br />
 
[11:19] danxor danx0r: minimap is too small to see individual dots<br />
 
[11:19] danxor danx0r: ahh zoom. nv<br />
 
[11:19] Tleiades Hax: danxor.. zoom works<br />
 
[11:19] danxor danx0r: mind<br />
 
[11:19] Neas Bade: at least that would be the goal<br />
 
[11:20] danxor danx0r: it's great to finally see what you all look like :-)<br />
 
[11:20] Tleiades Hax: lol<br />
 
[11:20] Michael Wright: I think charles is right about the map, I see my yellow dot for myself then 7 green dots (but there is 8 other people here) one person isn't getting a dot<br />
 
[11:20] Dalien1 Talbot: s/like/alike/ ? :)<br />
 
[11:20] Neas Bade: heh :)<br />
 
[11:20] danxor danx0r: sort of like that old "Addicted To Love" video<br />
 
[11:20] Charles Krinkeb: It builds a sense of community, and that is very good.<br />
 
[11:20] danxor danx0r: I see yello + 8<br />
 
[11:20] Dalien1 Talbot: hmm indeed me too.<br />
 
[11:21] Tleiades Hax: so, should I check the unit tests in?<br />
 
[11:21] Neas Bade: yes<br />
 
[11:21] Tleiades Hax: roger that<br />
 
[11:21] Neas Bade: how do you execute them at the moment?<br />
 
[11:21] Neas Bade: is it a manual run, or nunit?<br />
 
[11:22] Tleiades Hax: well, I set my unit test module as executable, and run "nunit-gui"<br />
 
[11:23] Tleiades Hax: we probably should set up a propper nunit test project<br />
 
[11:23] Neas Bade: lets start with what you have and work towards that goal<br />
 
[11:23] Tleiades Hax: ok<br />
 
[11:24] Tleiades Hax: having my tests in svn will actually help me, I have lost 'em twice already<br />
 
[11:24] Tleiades Hax: deleted the project folder, to get a fresh copy of opensim, and forgot to backup the tests<br />
 
[11:24] Dalien1 Talbot: given the speed with which the things change, i think it is good to push the stuff in as early as possible (assuming it does not break the build or functionality)<br />
 
[11:25] Dalien1 Talbot: even if it is piece by piece<br />
 
[11:25] Tleiades Hax: yes, small pieces at a time<br />
 
[11:25] Tleiades Hax: but I think some branches in svn would be a good thing<br />
 
[11:25] Dalien1 Talbot: hmm then you'd need to maintain more than one<br />
 
[11:25] Neas Bade: svn branches are actually a beast to merge<br />
 
[11:26] Michael Wright: we used to have branches as generally it just leads to problems<br />
 
[11:26] Dalien1 Talbot: i would rather try to keep it contained within one branch<br />
 
[11:26] Michael Wright: as= and<br />
 
[11:26] Neas Bade: I was thinking of creating a mercurial clone of svn trunk for people that wanted to do local branch changes<br />
 
[11:26] Charles Krinkeb: We declare a moratorium on updates from time to time for a few revisions if we need to pursue an idea that breaks the build for the general user.<br />
 
[11:27] Dalien1 Talbot: Charles: my build system always keeps the latest successful build<br />
 
[11:27] danxor danx0r: speaking of svn, I was thinking of creating opensim-libs<br />
 
[11:27] Neas Bade: in my experience using a distributed scm like mercurial makes life much easier if you want to do lots of branch stuff<br />
 
[11:27] Dalien1 Talbot: but ideally the build should never break<br />
 
[11:27] danxor danx0r: and starting down the path of getting a proper 3rd party build<br />
 
[11:27] Michael Wright: yeah, having a stable tag branch , that is only updated say once a week or whatever, but we tell everyone (non developers) to stick to that, is fine<br />
 
[11:27] danxor danx0r: what's mercurial? sounds like git<br />
 
[11:28] Neas Bade: danxor, good point<br />
 
[11:28] Charles Krinkeb: I was responding to Dan's comment a day or so ago that demanding the build always works may put the developers in a bind from time to time.<br />
 
[11:28] Neas Bade: danxor, yes, like git, but easier to learn to use<br />
 
[11:28] Dalien1 Talbot: MW: so that branch would be always re-synced automagically ? or needs manual maintenance ?<br />
 
[11:28] danxor danx0r: it'll need a merge every week sounds like<br />
 
[11:28] danxor danx0r: or maybe not<br />
 
[11:28] danxor danx0r: just replace it with the latest thing<br />
 
[11:28] Dalien1 Talbot: hmm that would mean manual maintenance... = pita.<br />
 
[11:29] danxor danx0r: hmmm<br />
 
[11:29] Dalien1 Talbot: could do simpler imho<br />
 
[11:29] Neas Bade: is this discussion on the stable tag, or on mercurial?<br />
 
[11:29] Dalien1 Talbot: just declare that the stuff should work on monday morning build or such<br />
 
[11:29] Tleiades Hax: stable tag, should be pretty easy to do, I think<br />
 
[11:29] danxor danx0r: if testing really works, you could build nightly, test, and update stable if test passes<br />
 
[11:30] Neas Bade: we need a few tests before we could believe that :)<br />
 
[11:30] danxor danx0r: good way to exercise the unit tests; if they miss something, we'll hear right away from users<br />
 
[11:30] Neas Bade: but that does tie nicely into test automation to move the stable branch<br />
 
[11:30] Dalien1 Talbot: danx0r: that's almost what i do on opensim.be, except i do not have the unittests - merely if the compile succeeds then i update with the new binary...<br />
 
[11:31] Neas Bade: http://opensimulator.org/wiki/Roadmap<br />
 
[11:31] Charles Krinkeb: Dalien, best to test that the client logs into the new binary to be sure.<br />
 
[11:31] Neas Bade: been trying to update that on the fly with things poping into my head<br />
 
[11:32] Dalien1 Talbot: ckrinke: indeed. once there are automated tests for that, I will include them.<br />
 
[11:32] Tleiades Hax: those automated tests, using libsl would be soooo cool<br />
 
[11:32] Dalien1 Talbot: although this boils down a bit to the unit testing<br />
 
[11:32] Dalien1 Talbot: i assume that every person that commits the code does test that the code actually works :)<br />
 
[11:32] Neas Bade: yes, I think automated tests with libsl will help a lot. I'm going to start digging into that tomorrow<br />
 
[11:33] Dalien1 Talbot: there were a couple of glitches - but they are usually caught by the build failures, it seems.<br />
 
[11:33] Tleiades Hax: the problem is testing on all platforms<br />
 
[11:33] Charles Krinkeb: I believe it is fair to say that code committed denotes responsibility to ensure users on both Linux & Windows have not gone backwards.<br />
 
[11:33] Dalien1 Talbot: ideally we should have an autobuild for both linux + windows<br />
 
[11:34] Neas Bade: right, well we need some automation there first, then we can see about that<br />
 
[11:34] Dalien1 Talbot: so that the windows folks did not have to bother with the linux setup, and vice versa.<br />
 
[11:34] Neas Bade: we've got opensimulator.org for the linux side<br />
 
[11:34] chill ken: is the stable tag idea due to time between releases being too long? what about doing more minor releases?<br />
 
[11:34] Dalien1 Talbot: neas: ah you started to do the builds there as well ?<br />
 
[11:34] Neas Bade: no, I haven't<br />
 
[11:34] Neas Bade: but we could<br />
 
[11:34] danxor danx0r: charles: not sure it's reasonable to expect every commit to include full windows + linux tests<br />
 
[11:34] danxor danx0r: unless they are automated<br />
 
[11:34] Dalien1 Talbot: ah ok - because http://ruth.opensim.be/build does do nightly build (nightly CET :)<br />
 
[11:35] danxor danx0r: I try to use my intuition about how far to test each commit<br />
 
[11:35] Neas Bade: yeh, and I think that's fair<br />
 
[11:35] Dalien1 Talbot: yeah i think it is a good approach.<br />
 
[11:35] Neas Bade: if people are running off svn they need to know there are risks there<br />
 
[11:35] Charles Krinkeb: No, but I would suggest it is reasonable that each commit, makes a commitment to ensure the users have moved forward and not backwards. That can be as simple as making sure at least one user compiles and runs on the other platform.<br />
 
[11:35] Tleiades Hax: I think almost all, run off svn head<br />
 
[11:35] danxor danx0r: that's exactly why there should be a semi-stable branch for the increasing number of 'power users' who compile but don't typically code<br />
 
[11:36] Dalien1 Talbot: Charles: I do not have the windows environment, and some folks do not have linux.<br />
 
[11:36] Neas Bade: well, I'm actually leaning against the stable tag, as it ends up being a crutch for not releasing more often<br />
 
[11:36] Charles Krinkeb: I would concur with Sean<br />
 
[11:36] Dalien1 Talbot: danx0r: maintaining more than one branch is a pain..<br />
 
[11:36] danxor danx0r: yeah, I think the defacto assumption has been, test n your own machine<br />
 
[11:36] Tleiades Hax: I'm leaning towards that point of view too<br />
 
[11:37] danxor danx0r: typically if you're not doing something involving build or native libs, you won't break the other platform<br />
 
[11:37] danxor danx0r: though I did have an exception where mono didn't run something .NET did<br />
 
[11:37] Tleiades Hax: the biggest pain is in the native libs<br />
 
[11:37] danxor danx0r: I suppose we could run mono on windows, and say mono is the gold standard for "this build doesn't segfault"<br />
 
[11:37] Neas Bade: I think that people are running on svn because the code is moving fast. We can't keep the code moving fast if we require a full regression test on every checkin :)<br />
 
[11:37] Dalien1 Talbot: Neas: indeed<br />
 
[11:38] danxor danx0r: Tleiades -- I want to talk about the lib problem<br />
 
[11:38] Neas Bade: yeh, native libs is a beast<br />
 
[11:38] Neas Bade: danxor: go for it<br />
 
[11:38] danxor danx0r: I think *all* libs, native or managed, that are not par tof standard build should be in a separate project<br />
 
[11:38] danxor danx0r: it will have two subfolders, managed & unmanaged<br />
 
[11:38] danxor danx0r: and each one has folders for each lib<br />
 
[11:39] danxor danx0r: then a top-level script to (eventually) invoke all the makes<br />
 
[11:39] danxor danx0r: (though some are VS hand-make only now)<br />
 
[11:39] Neas Bade: I like that approach<br />
 
[11:39] danxor danx0r: copy the dll's & so's into bin<br />
 
[11:39] danxor danx0r: and perhaps create zips -- tho that gets into architecture<br />
 
[11:39] Neas Bade: I think it would make sense not to polute bin/ with them though<br />
 
[11:39] Neas Bade: it would be nice to have a seperate lib/<br />
 
[11:39] danxor danx0r: ie, someone somewhere needs to build them for each target supported<br />
 
[11:40] danxor danx0r: neas (sdague?) -- dunno how mono .net deals with looking for them<br />
 
[11:40] danxor danx0r: but keep in mind stuff like ode.net.dll needs to find the unmanged ode.dll<br />
 
[11:40] danxor danx0r: AND, sometimes ode.dll will need to find other native stuff<br />
 
[11:40] Tleiades Hax: mono is actually easier than .net I think<br />
 
[11:40] Neas Bade: you just need to add paths in the prebuild.xml<br />
 
[11:40] Neas Bade: or nant<br />
 
[11:40] danxor danx0r: well I don't mind polluting bin, if it's not in subversion<br />
 
[11:41] danxor danx0r: ok I didn't explain right --<br />
 
[11:41] danxor danx0r: you can choose to build libs, _OR_ you just donwload a zip file<br />
 
[11:41] danxor danx0r: which someone built for your platform<br />
 
[11:41] danxor danx0r: and that zips into bin, or ./lib if you prefer<br />
 
[11:41] danxor danx0r: no binaries in subversion!<br />
 
[11:41] Neas Bade: well, what you are really talking about is opensim-libs releases<br />
 
[11:41] danxor danx0r: yes, right<br />
 
[11:41] Neas Bade: yep<br />
 
[11:42] Neas Bade: I'm all for that<br />
 
[11:42] danxor danx0r: but it's not going to be on the same schedule as opensim commits<br />
 
[11:42] Tleiades Hax: I like that too<br />
 
[11:42] Neas Bade: opensim-libs is an svn repo<br />
 
[11:42] danxor danx0r: yup<br />
 
[11:42] danxor danx0r: we can start with just reasonable file organization<br />
 
[11:42] Neas Bade: we can make it an svn external to the main opensim so you can have a single checkout to get the all if you are buliding from source<br />
 
[11:42] danxor danx0r: and the big makefile, all hail the make can come later<br />
 
[11:42] Neas Bade: well, probably we want nant instead of make<br />
 
[11:42] danxor danx0r: because for now, you gotsta do all the README's and INSTALL.txt<br />
 
[11:42] danxor danx0r: for each liv<br />
 
[11:43] danxor danx0r: lib<br />
 
[11:43] danxor danx0r: but at least, these will be the proper versions,<br />
 
[11:43] danxor danx0r: with maybe an opensim README about compile opts<br />
 
[11:43] Neas Bade: ok, danxor, you got the ball for that one?<br />
 
[11:43] danxor danx0r: so you can walk through it and end up at aknown state<br />
 
[11:43] Dalien1 Talbot: hm how do we "freeze" the version of the foreign lib ?<br />
 
[11:43] danxor danx0r: yeah (sigh)<br />
 
[11:43] Neas Bade: to do the new repo<br />
 
[11:43] danxor danx0r: I'll start it<br />
 
[11:43] Dalien1 Talbot: e.g. libsl<br />
 
[11:44] danxor danx0r: what else besides ode do I need to build?<br />
 
[11:44] danxor danx0r: libsl,<br />
 
[11:44] Neas Bade: libsl<br />
 
[11:44] danxor danx0r: is taht C++?<br />
 
[11:44] Dalien1 Talbot: if we just have it "pull the latest"<br />
 
[11:44] Neas Bade: it's libopenjpeg is the .so<br />
 
[11:44] danxor danx0r: did we get the version thing right now?<br />
 
[11:44] danxor danx0r: dalien -- "pull the latest" is a problem --<br />
 
[11:44] Dalien1 Talbot: then the changes in the libsl would break the opensim-libs<br />
 
[11:44] Dalien1 Talbot: precisely<br />
 
[11:44] Neas Bade: dalien, you can set a version number on an svn external IIRC<br />
 
[11:44] danxor danx0r: that's why we need copies of everything -- so we can say<br />
 
[11:44] danxor danx0r: we know this works with opensim<br />
 
[11:44] Neas Bade: sort of like a tag<br />
 
[11:44] danxor danx0r: you want to pull unstable and take a bet, go ahead<br />
 
[11:44] Dalien1 Talbot: or we svn co -r<br />
 
[11:45] Dalien1 Talbot: particular revision number<br />
 
[11:45] Neas Bade: http://svnbook.red-bean.com/en/1.0/ch07s03.html<br />
 
[11:45] Tleiades Hax: we are caught inbetween chairs on libsl<br />
 
[11:45] danxor danx0r: dalien -- well whoever maintains it needs to document what they do<br />
 
[11:45] danxor danx0r: there may be patches or subtle make changes<br />
 
[11:45] Dalien1 Talbot: cos we'd ideally then need the universal approach to all of the libs<br />
 
[11:45] danxor danx0r: for instance, I modified ODE's build to statically link those stupid msvc dll's<br />
 
[11:46] danxor danx0r: but at least it's there in svn, in a known state<br />
 
[11:46] Dalien1 Talbot: right... so could do something like gentoo & co - download the version that is needed, unpack, possibly patch, build ?<br />
 
[11:46] Tleiades Hax: msvc dll dependencies are always a pain<br />
 
[11:46] danxor danx0r: if someone does somethhing funky, yeah you just svn -r opensim-libs<br />
 
[11:46] danxor danx0r: nice<br />
 
[11:46] Stefan Andersson: Ok, so back.<br />
 
[11:46] Neas Bade: yeh, getting there would be great<br />
 
[11:46] danxor danx0r: tl -- turns out you can just compile statically but no one does<br />
 
[11:46] Stefan Andersson: Wow, quite the turnout. And no sim crashes.<br />
 
[11:46] danxor danx0r: ok I take the ball on this, with my copiouis free time<br />
 
[11:47] Tleiades Hax: yes, we do that at my workplace<br />
 
[11:47] Stefan Andersson waves YAY!<br />
 
[11:47] danxor danx0r: I'd love to have another commit-level dev on physics...<br />
 
[11:47] danxor danx0r: no crash because we all stand still with our arms out<br />
 
[11:47] Neas Bade: speaking of physics, how is that all going?<br />
 
[11:47] danxor danx0r: and act like it's IRC<br />
 
[11:47] danxor danx0r: physics update: got a major bug with terrain<br />
 
[11:47] danxor danx0r: strange one, hunting it down<br />
 
[11:47] Charles Krinkeb: Yep, you'all made a very stable sim server.<br />
 
[11:48] danxor danx0r: then some tweaks on bouncinetss<br />
 
[11:48] danxor danx0r: bounciness<br />
 
[11:48] paulie Femto: Howdy hi!<br />
 
[11:48] Neas Bade: hi paulie<br />
 
[11:48] danxor danx0r: darok made some bullet improvements but it needs real work<br />
 
[11:48] Tleiades Hax: hi paulie<br />
 
[11:48] danxor danx0r: Gerard did that awesome hollow patch, but it's one of 20+ prim types<br />
 
[11:48] danxor danx0r: all of which can be in combination -- so that's a ton of work<br />
 
[11:48] Neas Bade: are there specific things others can do to help out?<br />
 
[11:49] danxor danx0r: I'd like to pull GErard in if possible<br />
 
[11:49] danxor danx0r: well...<br />
 
[11:50] Tleiades Hax: at some point, work on assets will overlap with the prim types<br />
 
[11:50] Neas Bade: right, that's a good point tleiades<br />
 
[11:50] danxor danx0r: help: prim types<br />
 
[11:50] danxor danx0r: yes<br />
 
[11:50] danxor danx0r: it's some wicked stuff<br />
 
[11:51] danxor danx0r: common code to manage it would be great. Not sure who needs to know other than physics + client tho<br />
 
[11:51] danxor danx0r: scripts?<br />
 
[11:51] Tleiades Hax: scripts will need it<br />
 
[11:51] Tleiades Hax: to change prims based on scripts<br />
 
[11:51] Tleiades Hax: and those changes needs to be persisted into the asset store<br />
 
[11:52] Tleiades Hax: I scanned the ll functions today, and all of those functions are not implemented<br />
 
[11:52] Dalien1 Talbot: MW/lbsa: would like to spend 3-4 minutes with you later to discuss the TP stuff...(not to interrupt the physics thread now)<br />
 
[11:53] Charles Krinkeb: TL. Only about a third are implemented.<br />
 
[11:53] Neas Bade: right, I think actually getting assets somewhat more functional is probably required first<br />
 
[11:53] Tleiades Hax: yes, I noticed, and for pretty good reasons too<br />
 
[11:53] Neas Bade: which I'm very excited tleiades is digging into so much :)<br />
 
[11:53] Tleiades Hax: I plan on working on that, for a little while<br />
 
[11:53] Michael Wright: Dalien, okay but think for me at least, its going to have to be tomorrow, been a long day and need to spend some time resting<br />
 
[11:53] danxor danx0r: other physics issues:<br />
 
[11:54] danxor danx0r: in my opinion, ODE will not cut it without it being forked and worked on<br />
 
[11:54] danxor danx0r: it's just too damn unstable numerically<br />
 
[11:54] Dalien1 Talbot: ok.. actually i think i will need to discuss with Tleiades too :-)<br />
 
[11:54] danxor danx0r: it throws exceptions that need to be handled in C++<br />
 
[11:54] danxor danx0r: ppl have recommended OGre and other things<br />
 
[11:54] paulie Femto: Hows Bullet comin along?<br />
 
[11:54] Neas Bade: once we get assets working to a reasonable amount, there are a few other folks that I work with that are happy to grind out LSL functions<br />
 
[11:54] danxor danx0r: you sort of need a 'game engine' on top of physics, I've been ad-hacking it<br />
 
[11:55] danxor danx0r: bullet would need similar investment<br />
 
[11:55] danxor danx0r: but at least it's C#<br />
 
[11:55] paulie Femto: BulletX is more managed from the start, yeah.<br />
 
[11:55] Neas Bade: danxor, can you explain the difference between what we physics and game engine?<br />
 
[11:55] Neas Bade: in terms of what opensim needs?<br />
 
[11:56] danxor danx0r: well<br />
 
[11:56] danxor danx0r: phsics is very general<br />
 
[11:57] danxor danx0r: objects, vectors, momentum, collision<br />
 
[11:57] danxor danx0r: game engines add character classes<br />
 
[11:57] danxor danx0r: terrain<br />
 
[11:57] danxor danx0r: ability to modify & move stuff around without all hell breaking loose<br />
 
[11:57] danxor danx0r: the character class is probably the most important<br />
 
[11:57] danxor danx0r: and vehicle thingies<br />
 
[11:57] danxor danx0r: it can all be done from scratch, you know, just another layer of code to conceptualize and debug<br />
 
[11:58] danxor danx0r: I started out thinking it would be piece of cake<br />
 
[11:58] danxor danx0r: like everything, turns out it's sort of hard & stuff<br />
 
[11:58] Neas Bade: what do you think the best approach here is in terms of getting there sooner rather than later?<br />
 
[11:58] Neas Bade: :)<br />
 
[11:58] danxor danx0r: so if someone did it, I'd rather not have to<br />
 
[11:58] Neas Bade: what, it's not SMOP?<br />
 
[11:58] danxor danx0r: I think I need to look into what's available for C#<br />
 
[11:58] danxor danx0r: rather for .NET I guess<br />
 
[11:58] danxor danx0r: SMOP?<br />
 
[11:59] Neas Bade: simple matter of programming<br />
 
[11:59] danxor danx0r: sorry felt the need to collide<br />
 
[11:59] Tleiades Hax: no physics here<br />
 
[11:59] danxor danx0r: heh<br />
 
[11:59] Charles Krinkeb: basicphysics only on Sang<br />
 
[11:59] danxor danx0r: oh well<br />
 
[11:59] danxor danx0r: so can I walk thru you?<br />
 
[11:59] danxor danx0r: ok<br />
 
[12:00] danxor danx0r: hah SMOP exactly<br />
 
[12:00] Neas Bade: hehehe<br />
 
[12:00] danxor danx0r: frankly before this project, I poo-pooed "game engines"<br />
 
[12:00] Charles Krinkeb: "mere implementation detail"?<br />
 
[12:00] Tleiades Hax: can we break the physics thing into smaller pieces?<br />
 
[12:00] danxor danx0r: I figured I'm a macho programmer, etc<br />
 
[12:00] paulie Femto: Has anyone seen nixnerd around? Anyon eknow if he solved his issues with PHP launching a screen session?<br />
 
[12:00] paulie Femto: brb. door.<br />
 
[12:02] Charles Krinkeb: Sean: Do you envision this as a recurring weekly meeting and if so, where do we meet next Tuesday?<br />
 
[12:02] Neas Bade: ok, I need to run away for a bit and pick up food at the farm here. Can people look at the Roadmap I started at: http://opensimulator.org/wiki/Roadmap and embelish with more<br />
 
[12:02] Neas Bade: yes, I think we should try to do this every week<br />
 
[12:02] danxor danx0r: tleiades: one issue with physics is it's wired into the main code kinda slapdash<br />
 
[12:03] Neas Bade: given that we've all gotten a bit more scattered on IRC, it is helpful to have once a week of actually people all being roughly active at the same time<br />
 
[12:03] danxor danx0r: there was talk pre 0.4 of a proper event model<br />
 
[12:03] Charles Krinkeb: This sim is always up<br />
 
[12:03] Neas Bade: want to repropose event models on the -dev list?<br />
 
[12:03] Tleiades Hax listens<br />
 
[12:03] danxor danx0r: but we'd be more productive chatting if SL chat had name completion like IRC<br />
 
[12:04] Neas Bade: yeh, maybe :)<br />
 
[12:04] Dalien1 Talbot: danx0r: it's a client feature on IRC :)<br />
 
[12:04] Tleiades Hax: meeting here, is a good way of keeping track of our progress<br />
 
[12:04] Michael Wright: I think the tuesday meetings every week is good, its a little bit late for me, but we aren't going to get times that are good for everyone, and I guess one bonus of it being at this time on a tuesday is that if we stick to a hour, then Zero's office hour is straight after (ie its just starting now)<br />
 
[12:04] Neas Bade: yeh, that was also a sneaking idea by me :)<br />
 
[12:04] danxor danx0r: have LL said anything about physics in the WG?<br />
 
[12:04] danxor danx0r: it seems to get short shrift<br />
 
[12:04] Neas Bade: nope, not yet<br />
 
[12:05] danxor danx0r: like "yeah, we'll add some physics spice at the end"<br />
 
[12:05] Stefan Andersson: Yeah; it's a bit late for me too, I can't do any real work, but as MW said, it's good enough.<br />
 
[12:05] danxor danx0r: it shoud be a core ingredient, not a condiment :)<br />
 
[12:05] Charles Krinkeb: The best meetings have a beginning, a middle, and most important, an end.<br />
 
[12:05] Neas Bade: honestly, the lindens haven't been big on details for the AWG yet<br />
 
[12:05] Stefan Andersson: Alas, I need to go to bed, as I'm supposed to rise in like 5 hours.<br />
 
[12:05] Tleiades Hax: I am so totally at odds with the AWG<br />
 
[12:05] Stefan Andersson: Bye all, ace seeing you all on Sang.<br />
 
[12:05] Tleiades Hax: nite<br />
 
[12:05] Neas Bade: nite<br />
 
[12:05] danxor danx0r: well we need to do some wiki work on 0.5<br />
 
[12:05] Charles Krinkeb: nite all<br />
 
[12:05] Dalien1 Talbot: ok cya Stefan<br />
 
[12:06] danxor danx0r: are we all agreed that there be some sort of physics officially supported?<br />
 
[12:06] Stefan Andersson: (Sang's proabbly gonna be that legendary 'sirst island' that people talk about in SL)<br />
 
[12:06] Neas Bade: well, get over to SL now for Zero's office hours, and you can probably bring it up<br />
 
[12:06] danxor danx0r: ok<br />
 
[12:06] danxor danx0r: I'll probably just listen for a bit<br />
 
[12:06] Neas Bade: I took a screen shot<br />
 
[12:06] danxor danx0r: are we breaking up already?<br />
 
[12:06] Tleiades Hax: yes<br />
 
[12:06] Dalien1 Talbot: Neas: I'll ping you on SL for a TP ?<br />
 
[12:06] Dalien1 Talbot: I will blog the meeting :)<br />
 
[12:06] Neas Bade: I plan to do it every week, so we can have an animated progression of opensim<br />
 
[12:06] danxor danx0r: you look Mahvleous!<br />
 
[12:06] Neas Bade: catch you all later<br />
 
[12:07] danxor danx0r: later<br />
 
[12:07] Neas Bade: dalien, sure<br />
 
[12:07] Tleiades Hax: danxor, I think a lot will want to catch Zero Lindens office hours<br />
 
[12:07] danxor danx0r: where is zero's office<br />
 
[12:07] Tleiades Hax: hmm... I forgot<br />
 
[12:07] Dalien1 Talbot: danx0r: ping me on SL in a few - dalien talbot, once I get there via Sean, I TP you too :)<br />
 
[12:07] Dalien1 Talbot: cya on SL...<br />
 
[12:07] danxor danx0r: ok<br />
 
[12:08] Tleiades Hax: should we reconvene?<br />
 
[12:08] Dalien1 Talbot: lets continue in IM on the SL ?<br />
 
[12:08] Dalien1 Talbot: (and could do a group there, i suppose?)<br />
 
  
 
[[Category:Office Hour Logs]]
 
[[Category:Office Hour Logs]]

Latest revision as of 17:54, 27 October 2015

[10:45] Stefan Andersson: By the way; if anybody wants to explore the code, writing a simple 'module'
[10:45] Stefan Andersson: that logs all chat in a region to file would be a fun little project
[10:45] Dalien1 Talbot: indeed
[10:45] Tleiades Hax: and usefull for the meetings as well
[10:46] Stefan Andersson: MW introduced the module architecture so that people could write things more or less stand-alone.
[10:46] Dalien1 Talbot: shouldn't we better revive the IRC gateway module ?
[10:46] Dalien1 Talbot: that one was quite functional..
[10:46] Stefan Andersson: Personally, I'm surprised that we haven't had more people setting up custom sims and writing modules.
[10:46] Stefan Andersson: Why not have both?
[10:46] Dalien1 Talbot: true.
[10:47] Stefan Andersson: Everybody seems to want to hax core funcs.
[10:47] Tleiades Hax: the infrastructure for dynamically loading modules isn't in place yet
[10:47] Stefan Andersson: Which is fine; but I think the big breakthru comes the day people start pushing code from their own web sites.
[10:48] Stefan Andersson: Tlei, that's why I want people to get cracking at it. ;)
[10:48] Tleiades Hax: I think two things need to change, maybe three
[10:48] Tleiades Hax: we need a way do dynamically register plugins
[10:48] Tleiades Hax: we need to fill in the wiki page on modules
[10:49] Tleiades Hax: and as a possible last thing, but almost unrelated
[10:49] Stefan Andersson: That should be easy enough; a pick-up directory, or a part of the config.
[10:49] Tleiades Hax: I think we should isolate the plugin pattern we use into an interface
[10:49] Tleiades Hax: I think both
[10:49] Stefan Andersson: Well, yes; the plug-ins should have a well-defined facade into the system
[10:50] Stefan Andersson: that should coincide with the script engine facade
[10:50] Tleiades Hax: scan a subdirectory at run time, and look in the config file
[10:50] Stefan Andersson: As I said, simple enough, I think.
[10:50] Stefan Andersson: We already have dir-scanners and config readers..
[10:50] Tleiades Hax: yes, we have all the component parts
[10:51] Tleiades Hax: should be a piece of cake to do
[10:51] Stefan Andersson: Actually, we have so many of them they should be refactored into generic plug-in handlers
[10:52] Tleiades Hax: true
[10:52] Tleiades Hax: which is why I think we should define an IPlugin interface
[10:56] Stefan Andersson: Definitively
[10:56] Stefan Andersson: Actually, if we could get all our plug-ins into the same structure, so that the different plugin managers could just fetch a list of plug-ins implementing a certain type; that'd be like ace++
[10:57] Neas Bade: welcome all
[10:57] Neas Bade: we actually have a pretty reasonable turn out already
[10:57] Dalien1 Talbot: all of the plugins - like ?
[10:58] Stefan Andersson: So that the IPlugin should have something like IPlugin.Register( IRegisterPluginFacade facade )
[10:58] Dalien1 Talbot: ah something like meta-interface then ?
[10:58] Stefan Andersson: and IRegisterPluginFacade would have something like RegisterPlugin<TPlugin>( TPlugin instance )
[10:58] Neas Bade: man, it will be nice when we have chains and can sit :)
[10:59] Stefan Andersson: That would mean 'Register me as handling calls to interface TPlugin'
[10:59] Neas Bade: will someone volunteer to upload transcript of the office hours?
[11:00] Stefan Andersson: Then any Plugin could ask IPluginFacade for GetHandler<TPlugin>()
[11:00] Tleiades Hax: if I don't crash, I can do it
[11:00] Charles Krinkeb: I'll be the backup.
[11:00] Dalien1 Talbot: I'll also bounce it to the blog as well.
[11:00] Neas Bade: Tleiades, you have logging turned on?
[11:00] Stefan Andersson: Hm. GetHandler<TPlugin>( string name ) I guess
[11:00] Neas Bade: ok, cool
[11:00] Charles Krinkeb: is neas==sean?
[11:00] Neas Bade: yes
[11:00] Charles Krinkeb: got it
[11:00] Neas Bade is Sean Dague
[11:01] Neas Bade: here and in SL
[11:01] Michael Wright: seems like this is going to be everyone
[11:01] Neas Bade: sorry for showing up a little late, dalien / stefan, what did we I wonder into?
[11:01] Stefan Andersson: Um, anybody up for implementing what I just said? :P
[11:01] Neas Bade: yes, I agree
[11:02] Tleiades Hax: hmm. where do I turn on logging?
[11:02] Stefan Andersson: We were just discussing refining the plug-in handling.
[11:02] Stefan Andersson: Ok, so, I'm off for a while....
[11:02] Neas Bade: stefan: ok
[11:02] Tleiades Hax: k
[11:02] Charles Krinkeb: As the host, welcome to Sang. Which I believe to the most stable sim on osgrid. It is checked for functionality every morning, runs on the osgrid server and has not crashed once.
[11:03] Neas Bade: tleides it is under preferences somewhere
[11:04] Michael Wright: really need to fix up the time progressing sometime (night comes a bit too suddenly)
[11:04] Tleiades Hax: think of as being close to equator
[11:04] Neas Bade: Tleiades posted some suggested agenda on the wiki, which seems like as good a place to start as any. Though prior to that, are there other specific things people would like to talk about?
[11:04] Neas Bade: MW: heh :)
[11:05] Neas Bade: one of the things I'd like to throw out there is starting to build a master todo list on the wiki, as I think that will help letting people know where they can contribute bits
[11:06] Tleiades Hax: yes, agreed
[11:06] Neas Bade: man, everyone's quiet :)
[11:06] Dalien1 Talbot: i have crashed a couple of times :)
[11:06] Tleiades Hax: I think we should clean up the 0.5 list
[11:06] Tleiades Hax: make a free for all list of features they'd like
[11:06] Neas Bade: tleiades, I'd like to hear your thoughts on better approaches to the database layer as well
[11:07] Tleiades Hax: and a list of features we plan for 0.5
[11:07] Neas Bade: MW: no worries
[11:07] Tleiades Hax: I like the list the libsl guys have
[11:08] Neas Bade: url?
[11:08] Tleiades Hax: http://www.libsecondlife.org/wiki/Roadmap
[11:09] Tleiades Hax: as for database layer, I think we should consider things in stages
[11:09] Neas Bade: Ok, I just started a Roadmap page on the wiki
[11:09] Tleiades Hax: right now, focus should be on getting stuf to work
[11:09] Neas Bade: I'll take notes into it in "Uncategorized" for now and refine from there
[11:09] Tleiades Hax: kind of brute force
[11:10] Tleiades Hax: inventory is ok for mysql and sdague has sqlite well under control
[11:11] Michael Wright: the asset system is one area that so needs doing, I have been going to do it for weeks now, but as some of you know, for the last few weeks I've been very busy and just haven't had time
[11:11] Tleiades Hax: I have written some unit tests for that, and verified both sqlite and mysql
[11:11] Charles Krinkeb: Is it a correct assumption that inventory database stuff will not affect the sql users and regions table in grid mode?
[11:11] Neas Bade: tleides, well, I think you correctly pointed out that some of the use of DataSets ends up being problematic for larger data (as we load it all into memory)
[11:12] Tleiades Hax: I think mysql will be used for the larger grids, and sqlite for the standalone ones
[11:12] Neas Bade: I'm assuming you are going to take a slightly different approach on the mysql stuff for that?
[11:12] Tleiades Hax: that is my gut feeling, so no big worries I think
[11:12] Neas Bade: if you come up with a better pattern, I'll adapt the sqlite code to match that
[11:13] Tleiades Hax: yes, not quite as elegant as your code, but it won't flood server memory
[11:13] Neas Bade: yeh, that's fine :)
[11:13] Neas Bade: performance counts for something ;)
[11:13] Neas Bade: we can probably pull some patterns out of it and make it both pretty and performant
[11:14] Neas Bade: I'll admit to being pretty ado.net ignorant when I got started on that code, so learning as I go
[11:14] Tleiades Hax: IInventoryData implementation in MySql, is a sample, on how I plan to approach to MySQL
[11:14] Neas Bade: great
[11:14] Neas Bade: also, those unit tests you have, can you check them into the build tree somewhere sensible?
[11:15] Neas Bade: it would be really great to start getting tests in place for key function
[11:15] Tleiades Hax: well, problem with those are, what about programmers who aren't used to working with unit tests
[11:15] Charles Krinkeb: What sort of tests, neas?
[11:15] Neas Bade: we beat them with sticks until they like it :)
[11:15] danxor danx0r: heh
[11:16] Tleiades Hax: I have written some unitttests, for inventory data
[11:16] danxor danx0r: looks like yoga class
[11:16] Neas Bade: :)
[11:16] Charles Krinkeb: Perhaps tests could go on the wiki with simple step-by-step?
[11:16] nebadonb izumi: heheh yea
[11:16] Tleiades Hax: next will be unit tests for assets
[11:16] Michael Wright: yeah doing unit tests are always good.
[11:16] danxor danx0r: hi sorry I'm late -- technical glitches
[11:17] Neas Bade: danxor and I were actually talking about testing earlier, and I was thinking about starting to create a functional test suite by using libsl to create bots and walk them through the environment. It would help with isolating some of the region crossing bugs I think.
[11:18] Dalien1 Talbot: Neas: indeed. plus, with loadtesting as well...
[11:18] Charles Krinkeb: Hmm. 9 avatars, 8 green dots.
[11:18] Tleiades Hax: especially for the database plugins, it is the only way I can think of, which improves the probability of all the plugins working as expected
[11:18] Tleiades Hax: hey, that is a great idea, automated testing of sim functionality
[11:18] Tleiades Hax: I love that concept :-)
[11:18] danxor danx0r: unit tests, ftw
[11:18] Dalien1 Talbot: charles: self does not create the green dot :)
[11:18] Neas Bade: tleiades: agreed
[11:18] danxor danx0r: so would this be something we could all use on a single machine in stand-alone?
[11:18] nebadonb izumi: you dont see yourself charles
[11:19] Neas Bade: danxor, yes, I think so
[11:19] danxor danx0r: minimap is too small to see individual dots
[11:19] danxor danx0r: ahh zoom. nv
[11:19] Tleiades Hax: danxor.. zoom works
[11:19] danxor danx0r: mind
[11:19] Neas Bade: at least that would be the goal
[11:20] danxor danx0r: it's great to finally see what you all look like :-)
[11:20] Tleiades Hax: lol
[11:20] Michael Wright: I think charles is right about the map, I see my yellow dot for myself then 7 green dots (but there is 8 other people here) one person isn't getting a dot
[11:20] Dalien1 Talbot: s/like/alike/ ? :)
[11:20] Neas Bade: heh :)
[11:20] danxor danx0r: sort of like that old "Addicted To Love" video
[11:20] Charles Krinkeb: It builds a sense of community, and that is very good.
[11:20] danxor danx0r: I see yello + 8
[11:20] Dalien1 Talbot: hmm indeed me too.
[11:21] Tleiades Hax: so, should I check the unit tests in?
[11:21] Neas Bade: yes
[11:21] Tleiades Hax: roger that
[11:21] Neas Bade: how do you execute them at the moment?
[11:21] Neas Bade: is it a manual run, or nunit?
[11:22] Tleiades Hax: well, I set my unit test module as executable, and run "nunit-gui"
[11:23] Tleiades Hax: we probably should set up a propper nunit test project
[11:23] Neas Bade: lets start with what you have and work towards that goal
[11:23] Tleiades Hax: ok
[11:24] Tleiades Hax: having my tests in svn will actually help me, I have lost 'em twice already
[11:24] Tleiades Hax: deleted the project folder, to get a fresh copy of opensim, and forgot to backup the tests
[11:24] Dalien1 Talbot: given the speed with which the things change, i think it is good to push the stuff in as early as possible (assuming it does not break the build or functionality)
[11:25] Dalien1 Talbot: even if it is piece by piece
[11:25] Tleiades Hax: yes, small pieces at a time
[11:25] Tleiades Hax: but I think some branches in svn would be a good thing
[11:25] Dalien1 Talbot: hmm then you'd need to maintain more than one
[11:25] Neas Bade: svn branches are actually a beast to merge
[11:26] Michael Wright: we used to have branches as generally it just leads to problems
[11:26] Dalien1 Talbot: i would rather try to keep it contained within one branch
[11:26] Michael Wright: as= and
[11:26] Neas Bade: I was thinking of creating a mercurial clone of svn trunk for people that wanted to do local branch changes
[11:26] Charles Krinkeb: We declare a moratorium on updates from time to time for a few revisions if we need to pursue an idea that breaks the build for the general user.
[11:27] Dalien1 Talbot: Charles: my build system always keeps the latest successful build
[11:27] danxor danx0r: speaking of svn, I was thinking of creating opensim-libs
[11:27] Neas Bade: in my experience using a distributed scm like mercurial makes life much easier if you want to do lots of branch stuff
[11:27] Dalien1 Talbot: but ideally the build should never break
[11:27] danxor danx0r: and starting down the path of getting a proper 3rd party build
[11:27] Michael Wright: yeah, having a stable tag branch , that is only updated say once a week or whatever, but we tell everyone (non developers) to stick to that, is fine
[11:27] danxor danx0r: what's mercurial? sounds like git
[11:28] Neas Bade: danxor, good point
[11:28] Charles Krinkeb: I was responding to Dan's comment a day or so ago that demanding the build always works may put the developers in a bind from time to time.
[11:28] Neas Bade: danxor, yes, like git, but easier to learn to use
[11:28] Dalien1 Talbot: MW: so that branch would be always re-synced automagically ? or needs manual maintenance ?
[11:28] danxor danx0r: it'll need a merge every week sounds like
[11:28] danxor danx0r: or maybe not
[11:28] danxor danx0r: just replace it with the latest thing
[11:28] Dalien1 Talbot: hmm that would mean manual maintenance... = pita.
[11:29] danxor danx0r: hmmm
[11:29] Dalien1 Talbot: could do simpler imho
[11:29] Neas Bade: is this discussion on the stable tag, or on mercurial?
[11:29] Dalien1 Talbot: just declare that the stuff should work on monday morning build or such
[11:29] Tleiades Hax: stable tag, should be pretty easy to do, I think
[11:29] danxor danx0r: if testing really works, you could build nightly, test, and update stable if test passes
[11:30] Neas Bade: we need a few tests before we could believe that :)
[11:30] danxor danx0r: good way to exercise the unit tests; if they miss something, we'll hear right away from users
[11:30] Neas Bade: but that does tie nicely into test automation to move the stable branch
[11:30] Dalien1 Talbot: danx0r: that's almost what i do on opensim.be, except i do not have the unittests - merely if the compile succeeds then i update with the new binary...
[11:31] Neas Bade: http://opensimulator.org/wiki/Roadmap
[11:31] Charles Krinkeb: Dalien, best to test that the client logs into the new binary to be sure.
[11:31] Neas Bade: been trying to update that on the fly with things poping into my head
[11:32] Dalien1 Talbot: ckrinke: indeed. once there are automated tests for that, I will include them.
[11:32] Tleiades Hax: those automated tests, using libsl would be soooo cool
[11:32] Dalien1 Talbot: although this boils down a bit to the unit testing
[11:32] Dalien1 Talbot: i assume that every person that commits the code does test that the code actually works :)
[11:32] Neas Bade: yes, I think automated tests with libsl will help a lot. I'm going to start digging into that tomorrow
[11:33] Dalien1 Talbot: there were a couple of glitches - but they are usually caught by the build failures, it seems.
[11:33] Tleiades Hax: the problem is testing on all platforms
[11:33] Charles Krinkeb: I believe it is fair to say that code committed denotes responsibility to ensure users on both Linux & Windows have not gone backwards.
[11:33] Dalien1 Talbot: ideally we should have an autobuild for both linux + windows
[11:34] Neas Bade: right, well we need some automation there first, then we can see about that
[11:34] Dalien1 Talbot: so that the windows folks did not have to bother with the linux setup, and vice versa.
[11:34] Neas Bade: we've got opensimulator.org for the linux side
[11:34] chill ken: is the stable tag idea due to time between releases being too long? what about doing more minor releases?
[11:34] Dalien1 Talbot: neas: ah you started to do the builds there as well ?
[11:34] Neas Bade: no, I haven't
[11:34] Neas Bade: but we could
[11:34] danxor danx0r: charles: not sure it's reasonable to expect every commit to include full windows + linux tests
[11:34] danxor danx0r: unless they are automated
[11:34] Dalien1 Talbot: ah ok - because http://ruth.opensim.be/build does do nightly build (nightly CET :)
[11:35] danxor danx0r: I try to use my intuition about how far to test each commit
[11:35] Neas Bade: yeh, and I think that's fair
[11:35] Dalien1 Talbot: yeah i think it is a good approach.
[11:35] Neas Bade: if people are running off svn they need to know there are risks there
[11:35] Charles Krinkeb: No, but I would suggest it is reasonable that each commit, makes a commitment to ensure the users have moved forward and not backwards. That can be as simple as making sure at least one user compiles and runs on the other platform.
[11:35] Tleiades Hax: I think almost all, run off svn head
[11:35] danxor danx0r: that's exactly why there should be a semi-stable branch for the increasing number of 'power users' who compile but don't typically code
[11:36] Dalien1 Talbot: Charles: I do not have the windows environment, and some folks do not have linux.
[11:36] Neas Bade: well, I'm actually leaning against the stable tag, as it ends up being a crutch for not releasing more often
[11:36] Charles Krinkeb: I would concur with Sean
[11:36] Dalien1 Talbot: danx0r: maintaining more than one branch is a pain..
[11:36] danxor danx0r: yeah, I think the defacto assumption has been, test n your own machine
[11:36] Tleiades Hax: I'm leaning towards that point of view too
[11:37] danxor danx0r: typically if you're not doing something involving build or native libs, you won't break the other platform
[11:37] danxor danx0r: though I did have an exception where mono didn't run something .NET did
[11:37] Tleiades Hax: the biggest pain is in the native libs
[11:37] danxor danx0r: I suppose we could run mono on windows, and say mono is the gold standard for "this build doesn't segfault"
[11:37] Neas Bade: I think that people are running on svn because the code is moving fast. We can't keep the code moving fast if we require a full regression test on every checkin :)
[11:37] Dalien1 Talbot: Neas: indeed
[11:38] danxor danx0r: Tleiades -- I want to talk about the lib problem
[11:38] Neas Bade: yeh, native libs is a beast
[11:38] Neas Bade: danxor: go for it
[11:38] danxor danx0r: I think *all* libs, native or managed, that are not par tof standard build should be in a separate project
[11:38] danxor danx0r: it will have two subfolders, managed & unmanaged
[11:38] danxor danx0r: and each one has folders for each lib
[11:39] danxor danx0r: then a top-level script to (eventually) invoke all the makes
[11:39] danxor danx0r: (though some are VS hand-make only now)
[11:39] Neas Bade: I like that approach
[11:39] danxor danx0r: copy the dll's & so's into bin
[11:39] danxor danx0r: and perhaps create zips -- tho that gets into architecture
[11:39] Neas Bade: I think it would make sense not to polute bin/ with them though
[11:39] Neas Bade: it would be nice to have a seperate lib/
[11:39] danxor danx0r: ie, someone somewhere needs to build them for each target supported
[11:40] danxor danx0r: neas (sdague?) -- dunno how mono .net deals with looking for them
[11:40] danxor danx0r: but keep in mind stuff like ode.net.dll needs to find the unmanged ode.dll
[11:40] danxor danx0r: AND, sometimes ode.dll will need to find other native stuff
[11:40] Tleiades Hax: mono is actually easier than .net I think
[11:40] Neas Bade: you just need to add paths in the prebuild.xml
[11:40] Neas Bade: or nant
[11:40] danxor danx0r: well I don't mind polluting bin, if it's not in subversion
[11:41] danxor danx0r: ok I didn't explain right --
[11:41] danxor danx0r: you can choose to build libs, _OR_ you just donwload a zip file
[11:41] danxor danx0r: which someone built for your platform
[11:41] danxor danx0r: and that zips into bin, or ./lib if you prefer
[11:41] danxor danx0r: no binaries in subversion!
[11:41] Neas Bade: well, what you are really talking about is opensim-libs releases
[11:41] danxor danx0r: yes, right
[11:41] Neas Bade: yep
[11:42] Neas Bade: I'm all for that
[11:42] danxor danx0r: but it's not going to be on the same schedule as opensim commits
[11:42] Tleiades Hax: I like that too
[11:42] Neas Bade: opensim-libs is an svn repo
[11:42] danxor danx0r: yup
[11:42] danxor danx0r: we can start with just reasonable file organization
[11:42] Neas Bade: we can make it an svn external to the main opensim so you can have a single checkout to get the all if you are buliding from source
[11:42] danxor danx0r: and the big makefile, all hail the make can come later
[11:42] Neas Bade: well, probably we want nant instead of make
[11:42] danxor danx0r: because for now, you gotsta do all the README's and INSTALL.txt
[11:42] danxor danx0r: for each liv
[11:43] danxor danx0r: lib
[11:43] danxor danx0r: but at least, these will be the proper versions,
[11:43] danxor danx0r: with maybe an opensim README about compile opts
[11:43] Neas Bade: ok, danxor, you got the ball for that one?
[11:43] danxor danx0r: so you can walk through it and end up at aknown state
[11:43] Dalien1 Talbot: hm how do we "freeze" the version of the foreign lib ?
[11:43] danxor danx0r: yeah (sigh)
[11:43] Neas Bade: to do the new repo
[11:43] danxor danx0r: I'll start it
[11:43] Dalien1 Talbot: e.g. libsl
[11:44] danxor danx0r: what else besides ode do I need to build?
[11:44] danxor danx0r: libsl,
[11:44] Neas Bade: libsl
[11:44] danxor danx0r: is taht C++?
[11:44] Dalien1 Talbot: if we just have it "pull the latest"
[11:44] Neas Bade: it's libopenjpeg is the .so
[11:44] danxor danx0r: did we get the version thing right now?
[11:44] danxor danx0r: dalien -- "pull the latest" is a problem --
[11:44] Dalien1 Talbot: then the changes in the libsl would break the opensim-libs
[11:44] Dalien1 Talbot: precisely
[11:44] Neas Bade: dalien, you can set a version number on an svn external IIRC
[11:44] danxor danx0r: that's why we need copies of everything -- so we can say
[11:44] danxor danx0r: we know this works with opensim
[11:44] Neas Bade: sort of like a tag
[11:44] danxor danx0r: you want to pull unstable and take a bet, go ahead
[11:44] Dalien1 Talbot: or we svn co -r
[11:45] Dalien1 Talbot: particular revision number
[11:45] Neas Bade: http://svnbook.red-bean.com/en/1.0/ch07s03.html
[11:45] Tleiades Hax: we are caught inbetween chairs on libsl
[11:45] danxor danx0r: dalien -- well whoever maintains it needs to document what they do
[11:45] danxor danx0r: there may be patches or subtle make changes
[11:45] Dalien1 Talbot: cos we'd ideally then need the universal approach to all of the libs
[11:45] danxor danx0r: for instance, I modified ODE's build to statically link those stupid msvc dll's
[11:46] danxor danx0r: but at least it's there in svn, in a known state
[11:46] Dalien1 Talbot: right... so could do something like gentoo & co - download the version that is needed, unpack, possibly patch, build ?
[11:46] Tleiades Hax: msvc dll dependencies are always a pain
[11:46] danxor danx0r: if someone does somethhing funky, yeah you just svn -r opensim-libs
[11:46] danxor danx0r: nice
[11:46] Stefan Andersson: Ok, so back.
[11:46] Neas Bade: yeh, getting there would be great
[11:46] danxor danx0r: tl -- turns out you can just compile statically but no one does
[11:46] Stefan Andersson: Wow, quite the turnout. And no sim crashes.
[11:46] danxor danx0r: ok I take the ball on this, with my copiouis free time
[11:47] Tleiades Hax: yes, we do that at my workplace
[11:47] Stefan Andersson waves YAY!
[11:47] danxor danx0r: I'd love to have another commit-level dev on physics...
[11:47] danxor danx0r: no crash because we all stand still with our arms out
[11:47] Neas Bade: speaking of physics, how is that all going?
[11:47] danxor danx0r: and act like it's IRC
[11:47] danxor danx0r: physics update: got a major bug with terrain
[11:47] danxor danx0r: strange one, hunting it down
[11:47] Charles Krinkeb: Yep, you'all made a very stable sim server.
[11:48] danxor danx0r: then some tweaks on bouncinetss
[11:48] danxor danx0r: bounciness
[11:48] paulie Femto: Howdy hi!
[11:48] Neas Bade: hi paulie
[11:48] danxor danx0r: darok made some bullet improvements but it needs real work
[11:48] Tleiades Hax: hi paulie
[11:48] danxor danx0r: Gerard did that awesome hollow patch, but it's one of 20+ prim types
[11:48] danxor danx0r: all of which can be in combination -- so that's a ton of work
[11:48] Neas Bade: are there specific things others can do to help out?
[11:49] danxor danx0r: I'd like to pull GErard in if possible
[11:49] danxor danx0r: well...
[11:50] Tleiades Hax: at some point, work on assets will overlap with the prim types
[11:50] Neas Bade: right, that's a good point tleiades
[11:50] danxor danx0r: help: prim types
[11:50] danxor danx0r: yes
[11:50] danxor danx0r: it's some wicked stuff
[11:51] danxor danx0r: common code to manage it would be great. Not sure who needs to know other than physics + client tho
[11:51] danxor danx0r: scripts?
[11:51] Tleiades Hax: scripts will need it
[11:51] Tleiades Hax: to change prims based on scripts
[11:51] Tleiades Hax: and those changes needs to be persisted into the asset store
[11:52] Tleiades Hax: I scanned the ll functions today, and all of those functions are not implemented
[11:52] Dalien1 Talbot: MW/lbsa: would like to spend 3-4 minutes with you later to discuss the TP stuff...(not to interrupt the physics thread now)
[11:53] Charles Krinkeb: TL. Only about a third are implemented.
[11:53] Neas Bade: right, I think actually getting assets somewhat more functional is probably required first
[11:53] Tleiades Hax: yes, I noticed, and for pretty good reasons too
[11:53] Neas Bade: which I'm very excited tleiades is digging into so much :)
[11:53] Tleiades Hax: I plan on working on that, for a little while
[11:53] Michael Wright: Dalien, okay but think for me at least, its going to have to be tomorrow, been a long day and need to spend some time resting
[11:53] danxor danx0r: other physics issues:
[11:54] danxor danx0r: in my opinion, ODE will not cut it without it being forked and worked on
[11:54] danxor danx0r: it's just too damn unstable numerically
[11:54] Dalien1 Talbot: ok.. actually i think i will need to discuss with Tleiades too :-)
[11:54] danxor danx0r: it throws exceptions that need to be handled in C++
[11:54] danxor danx0r: ppl have recommended OGre and other things
[11:54] paulie Femto: Hows Bullet comin along?
[11:54] Neas Bade: once we get assets working to a reasonable amount, there are a few other folks that I work with that are happy to grind out LSL functions
[11:54] danxor danx0r: you sort of need a 'game engine' on top of physics, I've been ad-hacking it
[11:55] danxor danx0r: bullet would need similar investment
[11:55] danxor danx0r: but at least it's C#
[11:55] paulie Femto: BulletX is more managed from the start, yeah.
[11:55] Neas Bade: danxor, can you explain the difference between what we physics and game engine?
[11:55] Neas Bade: in terms of what opensim needs?
[11:56] danxor danx0r: well
[11:56] danxor danx0r: phsics is very general
[11:57] danxor danx0r: objects, vectors, momentum, collision
[11:57] danxor danx0r: game engines add character classes
[11:57] danxor danx0r: terrain
[11:57] danxor danx0r: ability to modify & move stuff around without all hell breaking loose
[11:57] danxor danx0r: the character class is probably the most important
[11:57] danxor danx0r: and vehicle thingies
[11:57] danxor danx0r: it can all be done from scratch, you know, just another layer of code to conceptualize and debug
[11:58] danxor danx0r: I started out thinking it would be piece of cake
[11:58] danxor danx0r: like everything, turns out it's sort of hard & stuff
[11:58] Neas Bade: what do you think the best approach here is in terms of getting there sooner rather than later?
[11:58] Neas Bade: :)
[11:58] danxor danx0r: so if someone did it, I'd rather not have to
[11:58] Neas Bade: what, it's not SMOP?
[11:58] danxor danx0r: I think I need to look into what's available for C#
[11:58] danxor danx0r: rather for .NET I guess
[11:58] danxor danx0r: SMOP?
[11:59] Neas Bade: simple matter of programming
[11:59] danxor danx0r: sorry felt the need to collide
[11:59] Tleiades Hax: no physics here
[11:59] danxor danx0r: heh
[11:59] Charles Krinkeb: basicphysics only on Sang
[11:59] danxor danx0r: oh well
[11:59] danxor danx0r: so can I walk thru you?
[11:59] danxor danx0r: ok
[12:00] danxor danx0r: hah SMOP exactly
[12:00] Neas Bade: hehehe
[12:00] danxor danx0r: frankly before this project, I poo-pooed "game engines"
[12:00] Charles Krinkeb: "mere implementation detail"?
[12:00] Tleiades Hax: can we break the physics thing into smaller pieces?
[12:00] danxor danx0r: I figured I'm a macho programmer, etc
[12:00] paulie Femto: Has anyone seen nixnerd around? Anyon eknow if he solved his issues with PHP launching a screen session?
[12:00] paulie Femto: brb. door.
[12:02] Charles Krinkeb: Sean: Do you envision this as a recurring weekly meeting and if so, where do we meet next Tuesday?
[12:02] Neas Bade: ok, I need to run away for a bit and pick up food at the farm here. Can people look at the Roadmap I started at: http://opensimulator.org/wiki/Roadmap and embelish with more
[12:02] Neas Bade: yes, I think we should try to do this every week
[12:02] danxor danx0r: tleiades: one issue with physics is it's wired into the main code kinda slapdash
[12:03] Neas Bade: given that we've all gotten a bit more scattered on IRC, it is helpful to have once a week of actually people all being roughly active at the same time
[12:03] danxor danx0r: there was talk pre 0.4 of a proper event model
[12:03] Charles Krinkeb: This sim is always up
[12:03] Neas Bade: want to repropose event models on the -dev list?
[12:03] Tleiades Hax listens
[12:03] danxor danx0r: but we'd be more productive chatting if SL chat had name completion like IRC
[12:04] Neas Bade: yeh, maybe :)
[12:04] Dalien1 Talbot: danx0r: it's a client feature on IRC :)
[12:04] Tleiades Hax: meeting here, is a good way of keeping track of our progress
[12:04] Michael Wright: I think the tuesday meetings every week is good, its a little bit late for me, but we aren't going to get times that are good for everyone, and I guess one bonus of it being at this time on a tuesday is that if we stick to a hour, then Zero's office hour is straight after (ie its just starting now)
[12:04] Neas Bade: yeh, that was also a sneaking idea by me :)
[12:04] danxor danx0r: have LL said anything about physics in the WG?
[12:04] danxor danx0r: it seems to get short shrift
[12:04] Neas Bade: nope, not yet
[12:05] danxor danx0r: like "yeah, we'll add some physics spice at the end"
[12:05] Stefan Andersson: Yeah; it's a bit late for me too, I can't do any real work, but as MW said, it's good enough.
[12:05] danxor danx0r: it shoud be a core ingredient, not a condiment :)
[12:05] Charles Krinkeb: The best meetings have a beginning, a middle, and most important, an end.
[12:05] Neas Bade: honestly, the lindens haven't been big on details for the AWG yet
[12:05] Stefan Andersson: Alas, I need to go to bed, as I'm supposed to rise in like 5 hours.
[12:05] Tleiades Hax: I am so totally at odds with the AWG
[12:05] Stefan Andersson: Bye all, ace seeing you all on Sang.
[12:05] Tleiades Hax: nite
[12:05] Neas Bade: nite
[12:05] danxor danx0r: well we need to do some wiki work on 0.5
[12:05] Charles Krinkeb: nite all
[12:05] Dalien1 Talbot: ok cya Stefan
[12:06] danxor danx0r: are we all agreed that there be some sort of physics officially supported?
[12:06] Stefan Andersson: (Sang's proabbly gonna be that legendary 'sirst island' that people talk about in SL)
[12:06] Neas Bade: well, get over to SL now for Zero's office hours, and you can probably bring it up
[12:06] danxor danx0r: ok
[12:06] danxor danx0r: I'll probably just listen for a bit
[12:06] Neas Bade: I took a screen shot
[12:06] danxor danx0r: are we breaking up already?
[12:06] Tleiades Hax: yes
[12:06] Dalien1 Talbot: Neas: I'll ping you on SL for a TP ?
[12:06] Dalien1 Talbot: I will blog the meeting :)
[12:06] Neas Bade: I plan to do it every week, so we can have an animated progression of opensim
[12:06] danxor danx0r: you look Mahvleous!
[12:06] Neas Bade: catch you all later
[12:07] danxor danx0r: later
[12:07] Neas Bade: dalien, sure
[12:07] Tleiades Hax: danxor, I think a lot will want to catch Zero Lindens office hours
[12:07] danxor danx0r: where is zero's office
[12:07] Tleiades Hax: hmm... I forgot
[12:07] Dalien1 Talbot: danx0r: ping me on SL in a few - dalien talbot, once I get there via Sean, I TP you too :)
[12:07] Dalien1 Talbot: cya on SL...
[12:07] danxor danx0r: ok
[12:08] Tleiades Hax: should we reconvene?
[12:08] Dalien1 Talbot: lets continue in IM on the SL ?
[12:08] Dalien1 Talbot: (and could do a group there, i suppose?)
Personal tools
General
About This Wiki