Present: hoijui, jK, Kloot, Tobi, zerver
- 83 Release will be very very likely this weekend
- ROAM, QTPFS & MultiTouch will follow in 84 which will be released early after 83
>> zerver joined!
>> Kloot joined!
>> Tobi joined!
>> jK joined!
jK: k still no release then :/
jK: hi all
ROAM now or later
jK: IHMO it's finished and I did my last commit to it
jK: even multithreaded it as much as I could
Kloot: yeah meetings could be more exciting lately
zerver: I have not tested it, and I think putting big changes in before the release is a bad idea
zerver: indeed meetings could be more exciting :)
jK: do we need hoij for a release?
zerver: we need a release now to get some action in the house
zerver: tobi is here to save us
Kloot: there is a release checklist for anyone who wants to practive
jK: what is with SF upload (don't think everyone with a SF account can do that) & linux package maintainers & minor: macosx app package
Tobi: you need to be in spring project as developer or so
jK: k I am
>> hoijui joined!
zerver: you came just in time for the release
jK: or should someone else do it?
hoijui: who will do it?
zerver: The checklist has some unix stuff so no for me I guess
hoijui: i should have my main machine back by tomorow
hoijui: i already have it back, but no internet there yet
hoijui: was on a netbook only, the last two weeks
hoijui: so ...
hoijui: i should be able to do it
hoijui: maybe not exactly tomrow, but one of these days
zerver: We are still on the first point of the agenda
hoijui: i wont have time to talk to eveyrone and all, like in the past, though
hoijui: i dont have it here
hoijui: someone pls pm link?
zerver: oh, have to erase that if we make minutes
jK: so if you do the release what's the ETA?
hoijui: i am quite out of touch here...
jK: k I will try so then
jK: do I have to inform the linux package maintainers before?
jK: or are they fast enough?
hoijui: never all fo them are fast enough
hoijui: if you can catch them before, tell them that it will happen.. if not, it will get to them somehow
jK: k ROAM & QTPFS will then be added in 84.0 following in <1month to 83.0
zerver: yes, that is the best imo
zerver: put them both in develop and I will test ROAM wrt MT in particular
hoijui: btw.. i think you have to do a release branch for this release too
hoijui: otherwise tagging will fail
jK: release and not master?
jK: so they should be equal?
hoijui: in the diagram, i made the first release (83.0) directly out of develop
jK: why will tagging fail?
hoijui: but you have to branch of a release branch first
hoijui: make at least one commit on that
hoijui: so... basically use the same schematics like we will use for all future releases
hoijui: becuase if you dont do that
hoijui: then the release commit will be on develop and on master
hoijui: so the release tag will be on develop too
hoijui: that is not allowed?
hoijui: has everyone updted changelog again?
hoijui: if not, just do that in the release branch
jK: I thought the release/master branch shouldn't contain any own commits (only cherrypicked ones)
hoijui: there should be no cherry picking anywhere
hoijui: master != release
hoijui: http://springrts.com/phpbb/viewtopic.ph ... 48#p501948
hoijui: see diagramm again
hoijui: release is a temp branch
hoijui: maybe i should do it? :D
hoijui: when i said i am out of touch, i meant mainly.. what is left to do.. but if you tell me nothign is left, i should be able to do it.. but not on fixed time
hoijui: somewhen this week
hoijui: but i am fine if not too
jK: k this weekend is release date, you will try to do it until then, else I will start Thursday
hoijui: meeting done?
jK: note: MultiTouch will be 0.84 candidate
jK: also regarding glShareList, it doesn't add a mutexes for each gl command (except ATi does something totally wrong)
jK: instead each threads/gl-context has it's own command queue and so gl commands are by design not threadsafe!
jK: (there are new mutex like objects available in OGL3 to make them threadsafe)
jK: so done :)
zerver: FPS @ 0.82 0.83
zerver: ST 77 73
zerver: MT 68 64 (55 with ShareLists)
zerver: these are my results with 1000 AK on screen, as you see 9 FPS drop with shared lists
zerver: ATI sucks maybe...
<< hoijui left!
jK: I don't doubt the slowdown, just the reason aren't mutexes
jK: switching command streams can be a time consuming act and depends a lot on good writen drivers
jK: also Fermi cards even are optimized for this case, by being able to process multiple command streams at the same time
zerver: yeah, pretty likely that this won't happen on NV, if someone feels like trying, set MultiThreadShareLists = 0 in config file