0.2.8.0_rc1: Release process and bugs

Website · Post by **Z-Man** » Thu Jan 12, 2006 12:27 am

Naah, I don't want to tag the sources yet. Every time I do that, there are some last minute changes.

Plan: Wait for Saturday. Let Philippe select preliminary art assets (Can I have the title screen already with the "Submit better art if you dare, deadline two weeks" message already? I hate image processing.) Tag sources. Build. Upload. Rejoice.

The CVS tag will be v0_2_8_0_rc1. The 0 comes in now because we may have a 0.2.8.1_rc1, but no further beta unless the sky falls on our heads.

Website · Post by **Z-Man** » Thu Jan 12, 2006 9:34 am

I'm now gone to the seminar and will be back Saturday afternoon. I'll try to catch two more bugs: the wrong arena size in the first round and the curious wall flickering you sometimes see.

Post by **Lucifer** » Fri Jan 13, 2006 1:06 am

From b0_2_8:

Code: Select all

Running automake...
src/Makefile.am:82: unused variable: `libtron_a_SOURCES'
src/Makefile.am:55: unused variable: `librender_a_SOURCES'
src/Makefile.am:92: unused variable: `libui_a_SOURCES'
src/Makefile.am:37: unused variable: `libengine_a_SOURCES'
Flagging scripts as executable...

Predictably, doesn't build. This is on Mandrake 10.1. I'll see if a source package builds.

Website · Post by **Z-Man** » Sat Jan 14, 2006 4:15 pm

That must be when only the master server is supposed to be built. I disabled the definition of the variables in this case.

Website · Post by **Z-Man** » Sun Jan 15, 2006 12:26 am

Sources are tagged. The new artwork is not yet included in CVS because it's preliminary and binary files spam the repository. I'll just copy it into the release folder for now.

Builds will be done and uploaded tomorrow, it's bedtime now.

Forgot to say: I got the two bugs I wanted to get during the seminar, and as a bonus, there is now a machine readable score log file. I'm working on a python program that turns that into a TRUE ladder, one that does not amount to "who can kill the most noobs". One where, in reasonable bounds, it does not matter how often or how long you play, just how well you play. Be sure to play on Sumo and kick some ass, I'll use that server for real world testing. Fortress logs will be used, too, but since we don't have a measure there yet on how you did in a single round (which is obviously required for a good ladder), the results will be less meaningful.

klax · Post by **klax** » Sun Jan 15, 2006 1:04 pm

I suppose we should wait to final artwork to upload builds.

Retaking the naming of .debs files... I plan to name this build this way:
Debian: armagetronad-0.2.7.1+0.2.8.0rc1-1.arch.deb
Ubuntu: armagetronad-0.2.7.1+0.2.8.0rc1-1ubuntu1.arch.deb
where arch will be i386 or amd64

Ubuntu naming should be only used if modifications are made against the original debian package. This is not the case but I'll use it to make ubuntu packages different from debian.
Any objections?

Website · Post by **Z-Man** » Sun Jan 15, 2006 3:11 pm

No, the new art assets won't be final in time. Only some of them made it, and they are included in the source tarball I'm desperately trying to push to SF just now. Since it isn't working so well, I'll attach it here, too.
Feel free to hack in the new rim wall texture, it seems like it is ready for inclusion. Or anything else you see and like.

The version tags look horrible, but if it is the best way to make them work, fine by me.

Website · Post by **Z-Man** » Sun Jan 15, 2006 3:45 pm

Sources, 32 Bit Windows and Linux autopackage and rpm builds are now up on sourceforge.

aabeta seems to be down, I'll PM luke.

Post by **dlh** » Sun Jan 15, 2006 4:20 pm

Something happened to font_s.png, it is blurry now.

Website · Post by **Z-Man** » Sun Jan 15, 2006 4:49 pm

Yeah, I noticed that, too, but did not find it too disturbing. It's wrtl's new version, feedback goes here:
http://forums.armagetronad.net/viewtopic.php?t=2889

Post by **dlh** » Sun Jan 15, 2006 5:33 pm

I am getting errors when I try to connect to my server -- "Waiting for real players (only spectators online)...". This seems to go on in an endless loop, shouldn't it give up after a little?

My server is running 0.2.8_alpha20060103

Website · Post by **Z-Man** » Sun Jan 15, 2006 7:15 pm

Whoops, sorry. Your server (and the dance of the cats ones) is running an intermediate CVS version, I thought Fortress was the only such server. The safe syncing code is incompatible with CVS servers between sometime after beta4.2 and rc1. Upgrade the server and everything should work fine.

What happens: the client thinks your server supports the "is syncing finished? syncing is finished" protocol, so it sends the "is sync finished?" query and waits for the response. That response does not come, so the client stays in the "syncing..." state for a while. There is a timeout eventually, but it's long: about 80 seconds. If you check the server logs, you'll notice me sitting through it twice.

Post by **dlh** » Sun Jan 15, 2006 7:58 pm

z-man wrote:Whoops, sorry. Your server (and the dance of the cats ones) is running an intermediate CVS version, I thought Fortress was the only such server. The safe syncing code is incompatible with CVS servers between sometime after beta4.2 and rc1. Upgrade the server and everything should work fine.

What happens: the client thinks your server supports the "is syncing finished? syncing is finished" protocol, so it sends the "is sync finished?" query and waits for the response. That response does not come, so the client stays in the "syncing..." state for a while. There is a timeout eventually, but it's long: about 80 seconds. If you check the server logs, you'll notice me sitting through it twice.

Alright, I'll upgrade soon. One problem -- I did the Mac OS X dedicated build, and I can't start it up.

Code: Select all

$ ./armagetronad-dedicated -v
Internal Error:Internal error in tConfItemBase::tConfItemBase(const char*) in /Users/me/armagetronad-0.2.8.0_rc1/MacOS/../src/tools/tConfiguration.cpp:161 : 
        Two tConfItems with the same name FLOOR_MIRROR_INT!
Please send a Bug report!
terminate called after throwing an instance of 'i'
Abort trap
$

Edit: Yep, that error is correct. FLOOR_MIRROR_INT is defined in both eDisplay and as a /dummy/ config item in tConfiguration

klax · Post by **klax** » Sun Jan 15, 2006 9:52 pm

Ubuntu Breezy 5.10 packages (debian maybe tomorrow...):

klax · Post by **klax** » Sun Jan 15, 2006 9:55 pm

Ubuntu Breezy 5.10 packages (amd64 ones, common files package is shared):