feature freeze

[Tobi]

As I pointed out in another thread, I'd like to release this year.

Therefore I'd like to request to focus on fixing bugs, increasing stability and getting the last major outstanding issues done (ie. player reordering in gamesetup).

Please wait a little with (huge) new features (or modifications) that may easily break stuff.

I'm confident we can do this.

Outstanding issues (out of my head, post here if you remember/know more):

• Stuttering/lag
• Gamesetup player/team number reordering
• Random start pos counts spectators too
• Memory corruption (I'm trying to track this down atm)

Of course there's also mantis, there are some major open issues there too:
http://spring.clan-sy.com/mantis/view_all_bug_page.php

Thank you!

[Kloot]

Peet reported that .take is broken (unverified
though, not enough people on the test server).

[LordMatt]

Someone join the test server, send me a PM and I'll test this with you.

[LordMatt]

Well, now spring segfaults on me when I try to load a script.txt (but not when I don't), so I can't test this. Running Ubuntu Feisty 64-bit, launching a separate x-server with custom xorg.conf. Compiling with scons from SVN r5124. The last working revision that I tested was before r5102.

Edit: http://pastebin.com/m573e02a0 contains the debug info from gdb.

Edit2: Fixed in r5126

[LordMatt]

[2:36:43 PM] <baczek> local replays are broken again
[2:43:20 PM] <[LCC]LordMatt> seems to cause spring to hang on loading with this in infolog.txt:
[2:43:37 PM] <[LCC]LordMatt> Section game\mapoptions missing in file buffer
[2:43:49 PM] <[LCC]LordMatt> Section game\modoptions missing in file buffer
[2:45:18 PM] <[LCC]LordMatt> same error for hosted replays
[2:46:49 PM] <[LCC]LordMatt> strangly when I end process it goes through the whole replay very fast (can hear it on the speakers)

[Kloot]

[2:46:49 PM] <[LCC]LordMatt> (...) it goes through the whole replay very fast (can hear it on the speakers)

r5012 seems to have introduced this one
(and r5128 did not fix it, non-gameSetup
demo's recorded by r5128 randomly either
get stuck in Waiting On Script mode or are
still played back in fast-forward).

[LordMatt]

Well that behavior happened only when I did end task on a frozen spring instance.

[AF]

Spring seems to freeze up when running at high simulation speeds using .setmaxspeed, and NTai causes a similar situation. The game continues inside the cpu, but no frames are rendered, and no user input is accepted. Debuggers and logs show this can go on for hours with entire AI games happening unseen.

[Auswaschbar]

There is a bugreport about it:
--> http://spring.clan-sy.com/mantis/view.php?id=727

[LordMatt]

redstar = AF (I think).

[AF]

Indeed it is so

[submarine]

i have encoutered that as well, happened a few times with large (2vs2) aai matches on higher game speed (+2 - +4). i thought it was caused by high cpu load, it has never happened in 1vs1 so far and it's not reproducable (with aai)

[AF]

It happens to me running at 3x speed.

[Tobi]

- Local replays are broken with r5138, I get
"Wrong connection ID in CNet::IsActiveConnection()" error when replaying demo recorded by client.

(I was sharing com with my Spring on my laptop vs two allied KAIK-0.13)

- Memory leak (temporary hard freeze, maybe requires hard reset on badly configured operating systems)
http://buildbot.no-ip.org/~tvo/spring/swap_peak.png
Kernel killed the process when it reached approx. 2G, which is all my physical RAM+swap filled

[Tobi]

- Immediate desync when comsharing vs 2 allied bots. Script.txt's seem correct:
client: http://pastebin.ca/833882
host: http://pastebin.ca/833885

[Tobi]

Peet reported that .take is broken (unverified
though, not enough people on the test server).

Did someone test this yesterday?

If so, does it work or is it still/again broken?


All other stuff in this thread is fixed AFAICS.
EDIT: though I don't know about the rendering freeze at high speed/load...

[Kloot]

LordMatt and I tried .take on wednesday with
~r5120, according to him it worked fine then
(although we only tested for one game ).

[Auswaschbar]

EDIT: though I don't know about the rendering freeze at high speed/load...

Should be fixed since r5141, but I can't test because I could not reproduce this bug.

[Tobi]

feature freeze over for those who don't follow the news page

[Auswaschbar]

So not wait until an upcoming bugfix release?