Planning for 0.77 Release

[Posts]

Is the dedicated server getting enough testing?(or any at all) it could end up having a significant impact on the future of spring and it would be tragic if we had to wait another year for the clients to be stable with it.

Last time i checked there was no publicly available autohost to test it with, just bibims unreleased autohost running on the main server. (don't see it right now, ~"spads")

[NOiZE]

Last I heard there is a massive issue with large games and spring freaking out.

Is there still a test server? Is it useful to test stuff? Is there some kind of list what needs to be tested?

[Hoi]

Last I heard there is a massive issue with large games and spring freaking out.

Is there still a test server? Is it useful to test stuff? Is there some kind of list what needs to be tested?

/facepalms

change server adress to buildbot.no-ip.org

[Masure]

Last I heard there is a massive issue with large games and spring freaking out.

Is there still a test server? Is it useful to test stuff? Is there some kind of list what needs to be tested?

0.77 Test lobbyserver: 82.239.138.44:8300

[Masure]

Last I heard there is a massive issue with large games and spring freaking out.

Is there still a test server? Is it useful to test stuff? Is there some kind of list what needs to be tested?

/facepalms

change server adress to buildbot.no-ip.org

I don't know if buildbot.no-ip redirects to the 0.77 test server.

[Beherith]

Ran a few test games, since the test server was empty; just some 4v4 ai vs ai runs, to check it out, see if it crashed. It didnt crash so far. Finally, decals can be turned on, which, imo is one of the prettiest little things and I missed it alot.

But I did find a small visual anomaly:
-Decals dont fade out abruptly, but fade out about halfway, then disapper suddenly. I recall in the older version that the decals setting just set the fade rate on the decals, so they faded away unnoticeably. Now, it seems there is a cap to how many there can be altogether, besides the fade timer. This makes it look wierd.

Unrelated: while testing ba 6.41 I noticed that it doesnt always draw health bars for wounded units, just when it feels like it.

[Crayfish]

Weren't the bulk of the problems to do with multiplayer sync issues? I can help test again if there's an opportunity.

[Masure]

Hi guys,

last evening I helped imbaczek in 0.77 testing (sync debug). I helped cause Satirik told me help was needed. If he didn't, I would have never thought there were so few people to help devs to test. I've been shocked that the lack of testers could be a brake to the devs debug willingness <- THAT SHOULD NEVER HAPPEN

1) The problem is people don't know when devs need help in testing.
2) Even if they know, they have to set up the test release.
3) Finaly they have to report things that devs need

Proposal :

1) Devs should make a new topic to announce their need as soon as they want to test something. No matter if this look like spam, helping the debug process is important. The location of those topics could be general discussion better than dev section cause less people read the dev section.

The best should be to also make an announce is the main lobby channel cause not all players read the forums.

One topic for one testing event with the date/time. The date/time could appear for each time zone to make it easier to read.

2) The topics should always contain the procedure to set up a spring folder for testing. This part could be copied/pasted for each topic because this won't change much between each tests.

eg :

a. Dupplicate your Spring folder and rename it to SpringTest
b. Download files :
- http:// ....
- http:// ....
- ...
c. Put file x1 in folder y1 (replace existing one)
Put file x2 in folder y2
...
d. Make a shortcut of the lobby exe and add a parameter to easily join the test server. Add -server ip:port

3) We should mention files that are supposed to be uploaded somewhere to help the debugging process. Add the url and the instructions




This looks obvious but I thing the procedure could dynamize the testing process. There is nothing worse than "no help in testing" when you re a dev.

Moreover you can join licho's or my TS server if this can make the testing session easier or friendlier.

Feel free to ask me if you need help in organization of the communication process.

edit : And just for the lulz, I just thought of a "Spring needs you for sync testing" image that could appear in the test event topics. A beautiful machio's paint could be awesome showing a commander with a hat pointing at you.

[reivanen]

+1 on that one.

I am more than willing to help out in testing, just make a topic about it and put it as welcome message in #main 24h before the test will take place so i can know about it.

The hope for a not-lagging cursor is alone enough of a reason to do anything i can in helping the release of new spring!

[dizekat]

+1 for the test threads. Currently theres not much way to help testing when you like build spring, login, see nobody here, logout, forget about it for a while then repeat cycle.

[el_matarife]

I am more than willing to help out in testing, just make a topic about it and put it as welcome message in #main 24h before the test will take place so i can know about it.

Also, please announce which SVN revision we will be using for the test so we don't end up having to install several times.

[Licho]

Btw you can also use springdownloader as distribution platform..
Make a module that auto keeps SVN version up-to-date and notifies user with some popup or sound that testing is needed.

[Patch]

I get some video error on big booms.

When i destroy the enemy carrier on gundam mod and it falls down, before crash to the ground the game is crash.

[ 6909] (0) F:\WINDOWS\system32\atioglxx.dll(atiPPHSN+0x3b2460) [0x6948B200]

I got that error with other mods. Radeon 2600 card, catalyst 8.8, WinXP
Version: spring_0.76b1+svn6448.exe

[mikedep333]

+1 For the test threads. I would be willing to help with them.

[zerver]

But I did find a small visual anomaly:
-Decals dont fade out abruptly, but fade out about halfway, then disapper suddenly. I recall in the older version that the decals setting just set the fade rate on the decals, so they faded away unnoticeably. Now, it seems there is a cap to how many there can be altogether, besides the fade timer. This makes it look wierd.

I noticed that trees show the same problem, but only if shadows are turned on. It appears to have something to do with the tree shadow fragment program.

If i change it like this in CAdvTreeDrawer::Draw, trees fade out like they should.

if(false && shadowHandler->useFPShadows){

The problem is easy to see if TreeRadius is reduced to a fairly low value. Someone with FP skillz should have a look at this...

[Kloot]

But I did find a small visual anomaly:
-Decals dont fade out abruptly, but fade out about halfway, then disapper suddenly. I recall in the older version that the decals setting just set the fade rate on the decals, so they faded away unnoticeably.

The decal fading process is now controlled by the GroundScarAlphaFade
config variable. By default it's set to 0 (no gradual fade-out), 1 gives the
old fading style.

[YokoZar]

Please hurry, not much time is left to jam this into Ubuntu 8.10 :)

[imbaczek]

getting there... reason for desync has been found, need a fix and some test games.

[clumsy_culhane]

Desync cause found! Yay! that should of been a news post.

[clericvash]

Desync cause found! Yay! that should of been a news post.

Hah yeah, good job on finding it though, i have seen quite a few logs on websvn of desyncs being fixed :D