Present: hoijui, jK(late), Kloot, Tobi, zerver
_Agenda_____________________________________________________________________
- Welcome
- Review meeting minutes past meeting
- Review 'how shall we keep track of who's doing what?'
- Progress of stuff to be done before release
- GPL (progress)
- dev statement thread progress (like for game content)
- OS X (progress)
- MT official build (progress)
- Password protection for connecting to game server
- Buildbot progress
- Next meeting
- Anything else? (WVTTK)
- End
_Main Conclusions___________________________________________________________
- Etherpad is good for keeping track of who's doing what
- Licensing an AI under a less restrictive license does not allow changing the license to something non-GPL compatible
- daftalx is invited for the next dev meeting
- new buildbot is up and running: http://springrts.com:7778/waterfall
generated files appear here: http://springrts.com/dl/buildbot/ - Lots of stuff left to do for the buildbot
- Create a Wiki list of Linux package maintainers, including contact info
_The meeting________________________________________________________________
Review meeting minutes past meeting
Tobi: so, nothing about the minutes from past week?
hoijui: i cant remember anything
Tobi: I suppose the format is well established now and there's nothing to comment on
Review 'how shall we keep track of who's doing what?'
Tobi: etherpad still working?
hoijui: .. again :D
zerver: the etherpad page is short and nice now
hoijui: mm
zerver: delete this point so we dont have to do it every meeting
hoijui: yep
hoijui: next
Tobi: ok, I can agree with that
Tobi: point removed
Progress of stuff to be done before release
Tobi: mostly jKs thing ofc but since I also have one blocking issue we can't just skip
hoijui: i guess that is the fat stuff on the todo list(s) ?
Tobi: I didn't fix it yet
Tobi: yeah
Tobi: oh and, zerver, did you intentionally move the malformed attack command bug to your to do list?
zerver: yes I took one of your blocking points tobi
Tobi: it's fine with me of course, but could have been someone accidentally dragged it
Tobi: ok, good
Tobi: do you have all the info for that bug?
zerver: because i stumbled across it by chance anyway and started fixing it
hoijui:
zerver: not really, is it in mantis?
hoijui: i just found out about the AIs still desyncing about two days ago
hoijui: will have to do syncdebug testing again
Kloot: hmm guess there is more pathfinder state involved then
Tobi: zerver, don't recall atm, but I can look at what I have (I think at least one piece of info in forum PM) and give it to you after meeting
hoijui: it woudl really help if BA would spawn its own commanders
zerver: The stuff I have fixed so far was complete lack of error handling in UnpackPacket
hoijui: yeah :/
Tobi: ok
zerver: so basically not only attack command could crash the game
Tobi: zerver: the crash I refered too was lack of checking in one of the *MoveTypes
Tobi: there is some if (params.size() == 3) {} else { /* assume params.size() == 4 */ } or something
...
zerver: anything else to fix before release?
Tobi: I'll add new stuff to my list in etherpad if I find things
Tobi: major thing I'll be working on is getting buildbot to make proper installers and stuff
hoijui: in mantis, we still have no tag for next release, right?
Tobi: since that is kind of blocking :)
zerver: very blocking
Tobi: (unless bibims one works fine now ofc)
Tobi: [ARP]hoijui_g5: I thought there was
* [RoX]Tobi checks
hoijui: was.. a problem?
hoijui: i though it is good now
Tobi: was.. tag for next release
hoijui: ah ok
Kloot: what about #1937? I've never seen it myself, but seems it could be blocking too
Tobi: there is 0.82 in the list
hoijui: maybe there were just too many :D
hoijui: ahh ok
Tobi: in advanced update you should be able to set a bug to that target version
hoijui: ok
hoijui: sorry then
Tobi: but since it's not marked as released you can not pick it when reporting a bug, I think
hoijui: mmm
hoijui: yeah maybe i wasjust lookign there
Tobi: anything else wrt progress for next release that needs discussing?
hoijui: not here
main points:
- jKs thing
- AI games desync (most likely fixed already by jK)
- malformed attack command (Tobi & zerver)
- getting buildbot to make proper installers & store (some) debug symbols
GPL (progress)
hoijui: i think we forgot to say who shoudl do anythign there :D
Tobi: I said I'd make a draft
hoijui: or was it clear that i had to do it, cuase i am the AI guy?
hoijui: ah ok
Tobi: but I didn't have much time for it yet, just started this morning on it but it isn't worth showing yet (only a few lines)
Tobi: or actually, this afternoon since in the morning I was still sleeping :)
Tobi: I suppose I'll try to finish the draft for next meeting...
Tobi: don't think there much else to say here
zerver: right, it does not impact next release at least so no big deal
Tobi: yeah
Kloot: about the "gpl or gpl compatible" clause: it should be made very clear that licensing under a less restrictive license should not allow changing the license to something non-gpl compatible :)
Tobi: agreed
zerver: yep
hoijui: ok
main points:
- Tobi makes a draft for the statement
- Licensing an AI under a less restrictive license does not allow changing the license to something non-GPL compatible
OS X (progress)
Tobi: points last time were subforum, VM, regular dev
Tobi: subforum is there now
zerver: good
hoijui: we should als aegis for the VM then?
Tobi: the one who requested it is also working on getting stuff compiling
hoijui: as we kind of have that dev...
Tobi: yeah
Tobi: at least, we don't know if he'll be 'regular' dev yet, but lets assume that for now :)
hoijui: yeah :D
Tobi: IMO, once he can get it compiling reliably through commandline commands, we can start working towards a build slave for OS X
hoijui: mmm
Tobi: he also said he himself possibly had some box for it
hoijui: ahh
Tobi: hmm suppose I should just ask him for the meeting next time, that makes it easier to discuss
hoijui: hmm
hoijui: or in the meeting thread for this week
Tobi: what do you mean?
zerver: minutes?
hoijui: the thread for the meeting minutes
hoijui: yep
Tobi: well I don't know if he reads those
Tobi: I'll just drop him a PM too
hoijui: ok, good
hoijui: next topic?
main points:
- ask aegis & daftalx for a permanently running OS X to be used as future build slave
- invite daftalx for next dev meeting
MT official build (progress)
zerver: it looks like hoi is doing good progress..
hoijui: yeah, separate build target now in cmake
hoijui: installer integration missing, and special handling by buildbot
Tobi: what kind of special handling will be needed?
Tobi: just invoking not only make but also 'make spring-mt' or so?
hoijui: yes exactly
Tobi: ok
hoijui: and only for some builds i guess
Tobi: so do the cmake options for MT still do something?
hoijui: no
hoijui: yes..
hoijui: GML_DEBUG
hoijui: but the other two not
Tobi: ok, so buildbot is now just compiling default build twice
hoijui: :D
Tobi: I'll change that then
hoijui: ok
zerver: and we need some code to pick up the mt exe and package it into the installer
hoijui: yeah, i can do that
hoijui: no big thing
zerver: i can assume it is easy, just copy/paste programming :)
hoijui: i guess we will have a section "Engine" then
hoijui: with default build always selected, and GML optional..
hoijui: default unselected?
zerver: not sure
hoijui: best woudl be, if we coudl check in nssi if the target system has multiple CPUs/Cores
hoijui: and default select if it has more then 1
zerver: yes
Tobi: since it will be built into the installer anyway I think it doesn't hurt to select it by default
Tobi: (i.e. it doesn't save any downloading while installing or so, only a tiny bit of disk space)
hoijui: hmm ok
zerver: there is risk few ppl will try it if it is optional only
** [LCC]jK joined the channel.
hoijui: yeah .. defautl select then
jK: hi, sorry I am a bit late
zerver: i think even headless should be installed by default
Tobi: and I can imagine checking for # of cores in nsis is not trivial
hoijui: hey jk
hoijui: headless by default, really?
Tobi: oh right it's headless too
hoijui: why do you think so?
hoijui: for more people to test it?
zerver: yes
hoijui: but...
Tobi: [LCC]jK: we are at MT official build (progress) atm
hoijui: i dont think we want the guys that do not even look at isntal options to test headless build udner windows
hoijui: hmm.. then agina.. htey usually wont as they will never see spring-hl.exe
Tobi: IMO headless is too special purpose to install by default
hoijui: ok .. default for both then :D
hoijui: .. then not
Tobi: same category as e.g. TASServer
zerver: you just need a spring-hl-readme.txt so they dont get confused why there is no gfx
hoijui: yeah.. will probably never be used under windows.. as windows alwyas has a GUI
hoijui: :D
Tobi: I wonder then though, is it even worth it to pack it in installer?
hoijui: hmm
Tobi: will make everyone download 5M more or so for maybe 1 or 2 people who will have a use for it
hoijui: yeah... i agree, dos not have ot be there
zerver: yeah, either dont pack it at all, or install it by default
hoijui: then the first
hoijui: is it possible to get it downloaded at isntall time?
zerver: that would be the best ofc, even for MT
hoijui: ie.. buildbot wise.. will ti be available on a URL?
zerver: to reduce DL size
Tobi: it shouldn't download directly from buildbot / springrts
hoijui: it is possible in nsis, for singel files
hoijui: hmm
Tobi: doesn't have any mirroring
Tobi: with SF you sit on a huge CDN
hoijui: ok..
hoijui: so we add option to DL it in the installer, but it will only work for reelases (which we put on SF) ?
hoijui: is ok for me
zerver: yeah
zerver: i think the main point is that if we are going to maintain headless, ppl must knot that it exists
hoijui: ..well... for GML, better integrate it into installer
hoijui: hmm
zerver: so some mention of it in the installer is necessary
zerver: yeah, probably integrate MT in installer for now
Tobi: what I don't like about installers that download stuff is that 1) installer can break if some server goes down / gives 404 on that URL and 2) you can't use it offline
zerver: if DL and mirroring can be set up reliably in the future we can change it though
zerver: i also hate downloading installers
hoijui: for me it is ok to not have headless in instalelr at all
hoijui: we shoudl add it to the dl page maybe
zerver: especially when the dowload turns out to be big
hoijui: and under linux, as separate package
hoijui: hmm
Tobi: oh and 3) every time you reinstall you need to redownload
zerver: thumbs down :)
Tobi: so, conclusion for now: integrate MT, default installed, don't integrate headless ?
zerver: fine by me
hoijui: yep
Tobi: maybe at some point we could also consider a big-everything you'd ever want installer and a normal one, mostly to update regular users (first install should come from game team's installer)
Tobi: but that's another point
hoijui: mmm
hoijui: next point
main points:
- separate CMake target for GML and Headless are done
- Tobi will adjust buildbot to build GML again
- GML will be integrated into installer (if built) and installed by default
Password protection for connecting to game server
Tobi: hmm
zerver: i think this point is done really
hoijui: i did nto change TASServer there so far
hoijui: ah..
Tobi: did this just remain on agenda since it was there last time accidentally?
hoijui: nothign needed?
zerver: oh tasserver, yes
hoijui: ok..
zerver: sry, not done then
hoijui: i have some major refactor going
hoijui: i was wokring on that...
hoijui: i shoudl put it aside and do this forst
hoijui: first
jK: so will it come with 82 or not?
zerver: we concluded this is not blocking iirc, the risk of spoofing is small
Tobi: and if it happens, it's just reason for someone to work more on lobby server :)
hoijui:
zerver: if someone has >15000ms lag then it can happen
hoijui: yeah
zerver: more importantly i will thoroughly test the reconnection code before release
Tobi: (I'll still try to set up lobby server peeps meeting btw, hoijui)
hoijui: ahh ok, good
Tobi: anything else for this point then?
main points:
- hoijui has to implement TASServer password per battle support (not bound to spring release)
- zerver will test the reconnection code before release
Buildbot progress
Tobi: it's green and it builds: http://springrts.com:7778/waterfall
hoijui:
Tobi: generated files now appear here: http://springrts.com/dl/buildbot/
Tobi: I didn't test any of them yet
hoijui: mm me neither
Tobi: since I use boost branch of mingwlibs I assume they don't work
hoijui: yeah :/
hoijui: i am setting up an ubuntu VM now
hoijui: to try to crosscompile there
Tobi: ok, cool
hoijui: with dwarf debug symbols
hoijui: well.. not really cool
hoijui: but am fed up with trying ti on gentoo
Tobi: well cool for us if you succeed <_<
hoijui: yeah
hoijui: for the buildbot...
Tobi: either way wrt buildbot I've still quite some thingies on my todo for it
hoijui: will there be a way to get to the builds more easily?
hoijui: ah ok
jK: what's the reason to recompile the boost libs btw?
Tobi: most important probably are to figure out some good way to trigger build
hoijui: casue the old ones fail wiht gcc 4.4+
Tobi: and at least storing debug symbols before release
hoijui: hmmm
Tobi: so we don't make a release for which we can never do address translation
hoijui: yeah
Tobi: of course also implementing the actual translation, but if necessary that could be done after release
hoijui: mm
hoijui: you can not store debug symbols of all builds?
Tobi: I can, but I didn't do that yet :)
hoijui: ok
Tobi: actually koshi already has a commit in a branch that does this, so possibly I just need to merge that and figure out a good place to store them
hoijui: yeah.. i guess.. does not make sense to ask for stuff as you still have lot to do there, and on your list already
hoijui: ahh ok
Tobi: so that's it I suppose
hoijui: yeah
main points:
- new buildbot is up and running: http://springrts.com:7778/waterfall
- generated files appear here: http://springrts.com/dl/buildbot/
- hoijui will try agian to cross-compile boost (MinGW GCC 4.4+ dwarf2)
- Tobi will work on debug symbol translation
- lots of stuff left to do for the buildbot
Next meeting
(same again, next week, 16:00 CET)
Anything else? (WVTTK)
hoijui: one thing..
hoijui: about linux packages
hoijui: it woudl be nice to have some more controll, or at least oversight over that
hoijui: who is doing them, adn hwo to contact them
Tobi: right
hoijui: maybe we make an etherpad for it
Tobi: it makes sense yeah because occasionally ppl e-mail me about ubuntu packages cause I happen to be owner of the LP repository
Tobi: but I don't even run Ubuntu atm
hoijui: and try to get all we know to "register" there? (name plus email/forum nic/lobby nic)
hoijui: hehe :D
hoijui: hmmm
Tobi: maybe wiki would be better for this
hoijui: ah yeah
hoijui: or first etherpad, adn then move to wiki.. the nthey dont have to register (in case soem are too lazy
)Tobi: well it doesn't hurt if one of us adds them to wiki
hoijui: yeah ok.. wiki only then
hoijui: zerver shoud ldo it, to get in contact with linux world
hoijui: :D
hoijui: nah i can do it
zerver: haha
Tobi: personally at some point I concluded that it's just a distro thing, and not upstream's job to bother about packages for all kinds of distros
Tobi: but I guess a small list of who maintains which doesn't hurt
hoijui: mm
Tobi: another random note: nice short meeting this time, almost within an hour :)
hoijui: yeah :D
zerver: actually i did lots it unix stuff at university so i'm not 100% M$
hoijui: ok
zerver: i even have unixutils installed so i can type "ls" and such in the command prompt
hoijui: heeh
hoijui: i meant more wiht the linux side fo spring
zerver: sure
jK: isn't that default installed with mysgit?
hoijui: hehe
Tobi: yeah, but you need to choose whether to make it available also to normal windows cmd
hoijui:
zerver: tobi, if you have anything more wrt the malformed attack bug, plz pm me
jK: I wonder if linux users will like the new mouse handling/feeling
Tobi: zerver: actually looking for it now
Tobi: zerver: baw maybe it's easier if I just fix it myself, I think that takes less time then trying to find the material on which I've based my conclusions till now (and since it wasn't in the code you fixed anyway, only slightly related..)
Tobi: zerver: so mind if I move it back? :)
hoijui: hehe
hoijui: learn to play together, kids
* [ARP]hoijui_g5 slaps himself
hoijui: doh
hoijui: one of you can play wiht the desync on AI games
hoijui: .. theres enough bugs for everyone
jK: any idea why?
jK: you said something of pathing?
hoijui: yeah
hoijui: as kloot said..
hoijui: the AIs can request paths from the engine
hoijui: this shoudl of course not change any state in the pathfinder
hoijui: but most likely it does
hoijui: it does so since the heat thing was interoduced
jK: I know heatmapping did something like that
hoijui: there were some attempts to fix it
jK: but I can't remember any changes (except my ones last week) to the pathfinder since 81
hoijui: kloot did some
hoijui: to fix it.. that was before your changes
hoijui: maybe he did fix it and your changes broke it again?
jK: I thought you mentioned those changes even earlier (than my changes)
jK: *desynced
jK: *desyncs
hoijui: hmm.. when were yoru changes..
hoijui: it was 2 or 3 days ago, that we got the desyncs
hoijui: or 4
Tobi: btw next time there is moderator meeting all of you are invited too
hoijui: ok
hoijui: i will check which rev it was..
hoijui: yeah, does include your changes
jK: I am not a moderator
hoijui: hehe :D
hoijui: i shoudl do tests..
hoijui: i hate to do desync tests :D
hoijui: but if the commit before yours will wokr.. we woudl at least know that its there
Tobi: [LCC]jK, doesn't matter, it's cause we are major stakeholders
Tobi: and ofc it's just an invite
Tobi: and besides there is no plan yet for another meeting :)
zerver: sry for being afk
zerver: yes tobi, maybe take your point back
main points:
- make a Wiki list of Linux package maintainers, including contact info
- nice short meeting this time, almost within an hour
- zerver is no unix noob
- jK just found the cause for AI desyncs <3
