Dev meeting minutes 2010-07-4

Dev meeting minutes 2010-07-4

Minutes of the meetings between Spring developers are archived here.
Post Reply
Spring Developer
Posts: 1865
Joined: 08 Oct 2006, 16:58

Dev meeting minutes 2010-07-4

Post by Kloot » 05 Jul 2010, 13:21

Date: Sunday, July 4, 2010 (2010-07-4)
Present: hoijui, jK, Tobi, Kloot

Agenda Main Conclusions
  • the RC will be released when all blocking bugs on the roadmap are fixed ("soon")
  • Spring will be converted to a github organization

[16:08:15] <[RoX]Tobi> == Progress of stuff to be done before release ==
[16:08:43] <[LCC]jK> I thought everything is ready
[16:08:47] <[RoX]Tobi> I just added 2 new items to the roadmap which people told me in chat:
[16:08:57] <Kloot> I thought there was going to be an RC last week ;)
[16:08:59] <[LCC]jK> but then there come ticket 1974
[16:09:32] <[ARP]hoijui_g5> yeah :D
[16:09:40] <[RoX]Tobi> Kloot: yeah that was plan, but unfortunately the time I had reserved for it was spent upgrading buildbot to newer mingw
[16:09:51] <[ARP]hoijui_g5> well.. RC yes, but curretn topic is for release
[16:09:53] <[LCC]jK> also 1970 was blocking
[16:10:17] <[ARP]hoijui_g5> valgrind log can be removed i guess
[16:10:20] <[LCC]jK> and I want to make today a RC, but there is ... 1974
[16:10:56] <[LCC]jK> I tested buildbot and those crash
[16:11:02] <Kloot> is 1974 related to 1969 maybe?
[16:11:03] <[LCC]jK> my own build ones don't
[16:11:55] <[RoX]Tobi> [LCC]jK: that was a build from ?
[16:12:05] <[LCC]jK> yup
[16:12:10] <[RoX]Tobi> ok
[16:12:12] <[LCC]jK> I copied my spring.exe in the dir and it worked fine
[16:12:19] <[LCC]jK> so it is nothing in a dll
[16:12:36] <[ARP]hoijui_g5> mingw version?
[16:12:59] <[RoX]Tobi> buildbot is 4.4.2
[16:13:04] <[LCC]jK> 4.4.0 here
[16:13:34] <[ARP]hoijui_g5> both are RELEASE builds?
[16:13:48] <[LCC]jK> here yes
[16:14:13] <[LCC]jK> I assume buildbot's `default` is RELEASE
[16:14:39] <[RoX]Tobi> it's cmake's default again iirc, which is RelWithDebInfo + stripping to separate debug infos in separate files
[16:14:41] <[ARP]hoijui_g5> WITHDEBINFO i guess
[16:15:09] <[ARP]hoijui_g5> (doh.. tricked myself)
[16:15:11] <[RoX]Tobi> bah well that needs some digging later on I suppose
[16:15:22] <[RoX]Tobi> I haven't tried it myself yet
[16:15:32] <[ARP]hoijui_g5> for the release (not RC) :
[16:15:40] <[ARP]hoijui_g5> there also seems to be a new issue wiht pathing
[16:15:55] <[ARP]hoijui_g5> quite often, consturctors do not move to the build site
[16:16:12] <[ARP]hoijui_g5> even if nothign is in the way
[16:16:26] <[ARP]hoijui_g5> i noticed it, and several others
[16:16:33] <[RoX]Tobi> is that in mantis?
[16:16:37] <[ARP]hoijui_g5> i will try to collect some images and demos..
[16:16:38] <[ARP]hoijui_g5> no
[16:16:44] <[RoX]Tobi> ok
[16:17:50] <[LCC]jK> how do I load debugsymbols of a separate file?
[16:18:50] <[RoX]Tobi> possibly symbol-file [ filename ], assuming you mean in gdb
[16:19:01] <Kloot> add-symbol-file iirc
[16:19:09] <[RoX]Tobi> ah or that yeah
[16:19:19] <[LCC]jK> works ty
[16:19:46] <[RoX]Tobi> anyway, apparently some things still to figure out for real rc
[16:20:13] <[LCC]jK> like?
[16:20:20] <[RoX]Tobi> keep putting issues you see in mantis and if they are blocking set them to target version 0.82 (in advanced update screen)
[16:20:25] <[RoX]Tobi> those that we just discussed :)
[16:20:43] <[RoX]Tobi> or more specifically:
[16:21:16] <[LCC]jK> security issue isn't blocking for RC
[16:21:42] <[RoX]Tobi> no, but may be for release
[16:22:17] <[RoX]Tobi> although I'd say it would be fine to just disable [...] and enable it once it is a bit more secure
[16:22:18] <Kloot> depends on how much time we allow between rc and release I guess
[16:23:38] <[RoX]Tobi> and the crash on start and quit, is that blocking for rc then?
[16:23:56] <[RoX]Tobi> personally I prefer if anything that's blocking for release is also blocking for rc, since after all that's why it's a *R* C
[16:24:05] <[ARP]hoijui_g5> yeah true
[16:24:22] <[ARP]hoijui_g5> i woudl say, this is blocking, yes
[16:24:52] <[ARP]hoijui_g5> next?
[16:24:55] <[LCC]jK> RC will have many bugs I assume, the earlier it comes the earlier we can fix the bugs and make a real release
[16:25:26] <[LCC]jK> so waiting for all bugs fixed is quite illusive
[16:25:33] <[RoX]Tobi> obviously
[16:25:37] <[RoX]Tobi> hence only blocking bugs :)
[16:26:20] <[RoX]Tobi> but sure, if everyone is fine with releasing latest buildbot build as is as RC then I can pick that up again
[16:26:38] <[RoX]Tobi> (apart from version number)
[16:26:51] <[RoX]Tobi> that would be version 0.82 then
[16:27:21] <[ARP]hoijui_g5> is ok for me
[16:27:36] <[RoX]Tobi> real release would then be 0.82.1 / .2 / .3 or so, probably
[16:27:42] <[ARP]hoijui_g5> but maybe try once more to fix
[16:28:13] <[LCC]jK> erm debugbuild isn't crashing
[16:28:28] <[LCC]jK> (from buildbot)
[16:28:51] <[RoX]Tobi> hmm it doesn't make those anymore for quite a while, you mean it doesn't crash in gdb?
[16:29:28] <[RoX]Tobi> or are you trying ?
[16:29:41] <[LCC]jK> somehow is find it quite strange if the public releases always are
[16:30:07] <[RoX]Tobi> yeah .rc would be better but would eat a bit of time to try how good that works
[16:30:53] <[LCC]jK> I tested 1175 (6/19)
[16:31:10] <[RoX]Tobi> ok
[16:31:31] <[RoX]Tobi> that is from before mingw upgrade
[16:31:52] <[RoX]Tobi> I upgraded it 6/28
[16:58:46] <[RoX]Tobi> mingw upgrade must have been 21st actually, now I think about it
[16:32:28] <[LCC]jK> k tested debugsymbols for newest then
[16:33:09] <[RoX]Tobi> anyway, shall we look into this sometime later?
[16:33:38] <[ARP]hoijui_g5> yeah
[16:34:06] <[LCC]jK> k
[16:34:10] <[RoX]Tobi> ok
[16:59:08] <[LCC]jK> btw
[16:59:09] <[LCC]jK>
[16:59:25] <[LCC]jK> looks like it's using SjLj again
[16:59:34] <[LCC]jK> it should use dwarf2
[16:59:53] <[RoX]Tobi> ok that could explain it I suppose
[17:00:44] <[RoX]Tobi> I can try again to recompile mingw, that's now a bit easier/safer now it's in a chroot

main points:
  • the RC was delayed/blocked because of Mantis #1974 (engine crashes on exit), which seems to be related to #1969
  • #1974 is not caused by external dll's, maybe related to exception format used by newer mingw version on buildbot
  • security issues surrounding #1975 are not considered RC-blocking, only release-blocking bugs are
  • Spring's version-name convention is not ideal for RC's

[16:34:11] <[RoX]Tobi> === GPL (progress) ===
[16:34:32] <[RoX]Tobi> I made the etherpad doc a little bit better this morning
[16:34:42] <[RoX]Tobi> I think it's good enough for posting
[16:35:20] <[RoX]Tobi> so feel free to review it or whatever, and if you all think it's good I make post
[16:38:31] <[ARP]hoijui_g5> it is good for me
[16:39:13] <[LCC]jK> for me, too
[16:39:17] <Kloot> +1
[16:39:26] <[RoX]Tobi> ok
[16:39:46] <[RoX]Tobi> then I'll post it asap
[16:40:09] <[ARP]hoijui_g5> :-)
[16:40:15] <[LCC]jK> and ty for the work
[16:40:40] <[ARP]hoijui_g5> .. yeah, thanks

main points:
  • developer stance on AI's wrt. the GPL is now finalized

[16:40:22] <[RoX]Tobi> next point: === security issues: ===
[16:40:37] <[RoX]Tobi> although I think this was mostly zerver's thing, isn't it?
[16:40:54] <[ARP]hoijui_g5> yeah
[16:41:15] <[RoX]Tobi> in very short, issue is that [...]
[16:41:49] <[ARP]hoijui_g5> even wiht the [...]?
[16:42:11] <[LCC]jK> specs don't need [...], when I understood it correctly
[16:42:14] <[RoX]Tobi> as I understood it, you don't need one now to [...]
[16:42:16] <[RoX]Tobi> yeah
[16:42:49] <[ARP]hoijui_g5> if we make an exception [...]?
[16:42:54] <[ARP]hoijui_g5> to always [...]
[16:43:30] <[RoX]Tobi> yeah that are solutions BD has suggested too
[16:43:44] <[RoX]Tobi> may work best to just do this discussion in the bug report and continue with next point
[16:43:52] <[RoX]Tobi>
[16:43:55] <[ARP]hoijui_g5> ok
[16:44:03] <[LCC]jK> normal players can have extended priviliges too (on springie etc.)
[16:44:14] <[RoX]Tobi> for now I just copy pasted the chat log to not forget completely about it
[16:44:18] <[LCC]jK> e.g. via !kick, !cheats
[16:44:27] <[ARP]hoijui_g5> mmm
[16:44:28] <[RoX]Tobi> yeah
[16:44:36] <[ARP]hoijui_g5> ok, no solution then
[16:44:49] <[ARP]hoijui_g5> maybe better discuss here, as i mantis.. it is slow
[16:45:05] <[LCC]jK> we would need zerver
[16:45:17] <[RoX]Tobi> yeah but we don't need a solution right at this moment, and indeed it would be more useful if zerver was here :)
[16:45:19] <[ARP]hoijui_g5> ah yeah true.. ok, next then

main points:
  • solutions to this issue will be investigated later

[16:45:29] <[RoX]Tobi> ==== * OS X (progress)=====
[16:45:54] <[RoX]Tobi> daftalx got a bit further
[16:46:02] <[RoX]Tobi> > Basically everything builds, what i'm getting now is a link error while looking for the "main" entry point for both spring-dedicated & spring-hl. I'm assuming there *is* a main function in these 2 targets ? Also there is an extra link error on spring-dedicated related to the ErrorMessageBox class.
[16:46:35] <[ARP]hoijui_g5> yeah, in HL it is even the same as in normal spring
[16:46:55] <[RoX]Tobi> for the errors see also here: ... logs/stdio
[16:47:11] <[RoX]Tobi> the one he means is:
[16:47:14] <[RoX]Tobi> Undefined symbols:
[16:47:15] <[RoX]Tobi> "_main", referenced from:
[16:47:15] <[RoX]Tobi> start in crt1.10.6.o
[16:47:15] <[RoX]Tobi> ld: symbol(s) not found
[16:47:34] <[RoX]Tobi> both when linking spring-dedicated and spring-hl
[16:47:36] <[ARP]hoijui_g5> i will have a lok.. no idea if i have a chance though
[16:47:41] <[ARP]hoijui_g5> look*
[16:47:55] <[LCC]jK> hmmm maybe some #defines?
[16:47:56] <[RoX]Tobi> could it be that SDL has some special tricks to make main() somehow 'more crossplatform' ?
[16:48:19] <[RoX]Tobi> and that maybe this is only half-present in spring-hl and spring-dedicated?
[16:49:21] <[ARP]hoijui_g5> yes, both of these
[16:49:25] <[ARP]hoijui_g5> quite likely
[16:49:28] <[ARP]hoijui_g5> combined
[16:49:56] <[RoX]Tobi> ok
[16:50:20] <Kloot> looks like main.cpp fails to compile at all
[16:51:15] <[RoX]Tobi> as in, you think the warnings are related?
[16:51:52] <Kloot> yeah, as though main() is not visible/exported
[16:52:19] <[RoX]Tobi> hmm that might be an explanation, wonder why that would be though
[16:52:38] <Kloot> but the linker seems to want symbol "_main"
[16:53:16] <[ARP]hoijui_g5> yeah, thats most likely cuase of SDL
[16:53:21] <[ARP]hoijui_g5> redefining stuff
[16:53:58] <Kloot> could check SDL source to see what it does exactly
[16:54:47] <[ARP]hoijui_g5> yeah.. i did that some 3 weeks ago
[16:54:57] <[ARP]hoijui_g5> remember something like that
[16:55:02] <[ARP]hoijui_g5> its al in hte headers
[16:55:43] <[RoX]Tobi> ok
[16:56:00] <[LCC]jK> next?
[16:56:02] <[RoX]Tobi> well if you have some clue how to fix please try or you could suggest something to try to daftalx :)

main points:
  • Spring fails to link on OSX buildbot, likely due to SDL redefining main()

[16:56:19] <[RoX]Tobi> === * MT / LUA (silently rename mt -> gml) ===
[16:57:08] <[LCC]jK> hoijui, you want to reverse the rename?
[16:57:09] <[ARP]hoijui_g5> yeah i though.. as zerver is not here.. quick quick.. ;-)
[16:57:13] <[ARP]hoijui_g5> hehe
[16:57:19] <[ARP]hoijui_g5> its a joke
[16:57:28] <[LCC]jK> ^^
[16:57:28] <[RoX]Tobi> k lol :p
[16:57:30] <[ARP]hoijui_g5> but... i still think we should have used gml eveyrwhere
[16:58:00] <[ARP]hoijui_g5> as zerver is not here, we can skip this
[16:58:05] <[RoX]Tobi> ok

main points:
  • hoijui is not happy with the s/GML/MT/g change for the GML-multithreaded binary

[16:58:11] <[RoX]Tobi> ==== * Buildbot progress ======
[16:58:36] <[RoX]Tobi> nothing of importance since past week IIRC

main points:
  • none

[17:01:55] <[RoX]Tobi> === GitHub: convert spring into an organization? ===
[17:02:01] <[RoX]Tobi> * see:
[17:02:15] <[RoX]Tobi> +1 from me
[17:02:59] <[ARP]hoijui_g5> ok
[17:03:17] <[ARP]hoijui_g5> for me too
[17:04:15] <[LCC]jK> +1
[17:04:50] <[ARP]hoijui_g5> kloot will not answer, he wants to keep it short
[17:05:00] <[ARP]hoijui_g5> next
[17:05:07] <Kloot> any lines I can save ^^
[17:05:11] <[ARP]hoijui_g5> :D
[17:05:48] <Kloot> but yeah, organization seems natural with all the repo's spring now has
[17:06:11] <[RoX]Tobi> who will upgrade it?
[17:06:17] <[ARP]hoijui_g5> me mee
[17:06:19] <[RoX]Tobi> k

main points:
  • Spring will be converted to a github organization for easier project management, etc

[17:06:30] <[RoX]Tobi> === Next meeting ===
... (everyone voted +1 on the same time next week)

[17:10:25] <[RoX]Tobi> ===Anything else? (WVTTK)===
[17:11:00] <[ARP]hoijui_g5> i tried to translate a stack trace with an AI crash, and it did nto translate the AI part
[17:11:05] <[ARP]hoijui_g5> that was wiht RAI
[17:11:08] <[ARP]hoijui_g5> (new buildbot)
[17:11:23] <[ARP]hoijui_g5> did it work for you, tobi?
[17:11:26] <[LCC]jK> possibly SjLj problem
[17:11:57] <[ARP]hoijui_g5> ahh yeah could be
[17:12:06] <[ARP]hoijui_g5> old buildbto had similar problems, form time to time
[17:13:05] <[RoX]Tobi> translation shouldn't be related to SjLj I think
[17:13:21] <[ARP]hoijui_g5> .. also true :D
[17:13:29] <[LCC]jK> SjLj can't catch errors in dlls afaik
[17:13:43] <[ARP]hoijui_g5> yeah but if ther is a stack trace ..
[17:13:53] <[ARP]hoijui_g5> exceptions do not matter anymore, right?
[17:16:17] <[RoX]Tobi> correct
[17:16:31] <[RoX]Tobi> although I don't know what kind of weird effects that signal handler that triggers a segfault has

main points:
  • stacktrace translation for AI's is unreliable, but should not be affected by the exception format used
0 x

Spring Developer
Posts: 1358
Joined: 16 Dec 2006, 20:59

Re: Dev meeting minutes 2010-07-4

Post by zerver » 05 Jul 2010, 15:34

Code: Select all

main points:

    * hoijui is not happy with the s/GML/MT/g change for the GML-multithreaded binary
The main reason I want the "MT" name is that (large parts of) GML will likely be removed/disabled when we introduce glShareLists. And that will happen soon, not this release, but probably the next. I don't think it is a good idea to change the name back and forth. Better to settle for a more generic name that people can understand already now.
0 x

User avatar
Former Engine Dev
Posts: 4342
Joined: 22 Sep 2007, 09:51

Re: Dev meeting minutes 2010-07-4

Post by hoijui » 05 Jul 2010, 17:49

i do not understand. how is that a reason for the name MT instead of GML?

for me..

pro MT:
people know what MT means, not so with GML (-> more people will use it)
contra MT:
it gives a wrong idea of what it is. as people think:
"ahh, this will use all my 4 cores to the max, and make the game run much smoother. i can play with unit limit 2000, and let 16 AIs play against each other on a 32x32 map"
MT should be reserved for when we have a real MT build (eg, Sim is MT). independent of whether it will ever exist or not.

you had a lot of work with it, and it would be mean to rename it back. i do not plan to do that ;-)
but i believe it will back-fire, at least a bit.
0 x

Engines Of War Developer
Posts: 583
Joined: 09 Jun 2005, 23:39

Re: Dev meeting minutes 2010-07-4

Post by SeanHeron » 05 Jul 2010, 21:27

This is an outside opinion (of course), but I'd doubt whether it's named Multi-threaded or whatever GML stands for will make as much difference as you seem to think Hoijui - the real question is which is default, and (if no) how easy an option is it to set/use the MT/GML version. As I'd be surprised if it were made default as long as it still has the given difficulties; and I've never tripped over the option on installation yet, I don't think you have a real issue.

That doesn't mean that noone is going to use it in a way thats detrimental too himself, and not realise why thats the case, but I think it will be very few - and if the installation route is even via choosing a different download (just looked at the download page, and that seems to be the case), then I think it is absolutely no-worries (esp. given the prominent warning there).
0 x

Posts: 834
Joined: 19 May 2009, 21:10

Re: Dev meeting minutes 2010-07-4

Post by SirMaverick » 05 Jul 2010, 21:56

Is using passwords a solution?
0 x

Spring Developer
Posts: 4598
Joined: 01 Jun 2005, 11:36

Re: Dev meeting minutes 2010-07-4

Post by Tobi » 05 Jul 2010, 22:00

That is being discussed :-)

Unrelated, I can't make RC until at best friday since I'm in Lyon for work from tomorrow till thursday. Don't think I'll do lots of Spring things there, let alone making an RC :-)
0 x

User avatar
Former Engine Dev
Posts: 4342
Joined: 22 Sep 2007, 09:51

Re: Dev meeting minutes 2010-07-4

Post by hoijui » 05 Jul 2010, 22:57

eyy Lyon! i will be near there in 4 weeks... nice! :D

MT spring will be in next releases installer, and using it will be as easy as switching the executable name in your lobby.
and even now, where it is not yet available that easily, many community insiders that try it get wrong perceptions of it, like the ones i described.
i know it is a problem, question is just, how much it will scale up when it is in the installer, and how much rants we will get in the help & bugs forum. well.. after the first few, we'll just have to redirect them to the other threads ;-)
0 x

Spring Developer
Posts: 4598
Joined: 01 Jun 2005, 11:36

Re: Dev meeting minutes 2010-07-4

Post by Tobi » 11 Jul 2010, 12:09

Not entirely sure I can be properly online during today's meeting, but I'll try :-)

Just start at the normal time though, if I don't join there were some troubles getting online ;-)
0 x

User avatar
Spring Developer
Posts: 2299
Joined: 28 Jun 2007, 07:30

Re: Dev meeting minutes 2010-07-4

Post by jK » 11 Jul 2010, 14:52

Meh me too.

Still I hope to be there between 16.30-17.00.
0 x

User avatar
Former Engine Dev
Posts: 4342
Joined: 22 Sep 2007, 09:51

Re: Dev meeting minutes 2010-07-4

Post by hoijui » 11 Jul 2010, 23:36

there was no meeting today (11. July 2010).
0 x

Post Reply

Return to “Meeting Minutes”