Release process for 0.2.7.1

Post by **Tank Program** » Sun Mar 06, 2005 7:11 pm

Wellllll... Looks like you've got everything under control, and as my rc6 installer seems to be a marked failure, I guess I'll leave it to you to take care of klax.

Website · Post by **Z-Man** » Sun Mar 06, 2005 7:12 pm

Another bug in the rc6-installer: The uninstaller does not remove the start menu entry. Probably just a conversion error that won't happen in the final release, but anyway.

Here come the sources of rc8. Changes:
- Reverted to old armagetron.dsp, only changed the name of the generated executable to armagetronad.exe
- Fixed the missing 0 termination in uMenu.cpp

Since the bug in uMenu.cpp is VERY old and got by unnoticed, new builds are not strictly required for all non-Windows operating systems.

To clarify the .dsp situation: All previous DSPs from source archives of RCs are broken. Only the old ones from CVS revision 1.2 or the ones in this archive work.

What new files have you added? When I compiled, I did not add them so I am not seeing this bug.

I did not add files, sorry if that was inaccurate. I changed the windows visual studio build files to a theoretically better setup, but it backfired. Linux and OSX should be unaffected.

The language files and docs from the Windows source archive should be safe to use. Nothing like "Advanced Advanced"

Edit: the .nsi scripts did not get converted to Windows newlines. NSIS does not seem to mind.

klax · Post by **klax** » Sun Mar 06, 2005 7:46 pm

Windows Installers of RC8.
Edit: removed. Try z-man ones.

I believe this is the 1st program I see with a RC8 lol

tank: I have to build the installers severals times because I always forget to update the exes, doc, languages, uff. With an automated process we won't have these problems.

Website · Post by **Z-Man** » Sun Mar 06, 2005 7:48 pm

Windows binaries ( they'll install into the default location over your previous installation, so take care! ). They are unofficial, Tank will want to build new ones later. Heck, everything is unofficial here anyway

And the client RPM.

Website · Post by **Z-Man** » Sun Mar 06, 2005 7:57 pm

Sorry for the dublicate work...

klax wrote:tank: I have to build the installers severals times because I always forget to update the exes, doc, languages, uff. With an automated process we won't have these problems.

That's what the makedist.bat is for. It'll be extended to fully work from the CVS sources. All that is missing is m4.

Attaching rest of rpms...

n54 · Post by **n54** » Sun Mar 06, 2005 9:04 pm

wow "release quickly and release often" indeed

great work, so far rc6 was the best imo - i'll get rc8 now

edit: umm err i meant rc5 of course

Post by **Tank Program** » Sun Mar 06, 2005 9:42 pm

z-man wrote:Tank will want to build new ones later.

Build binaries? Sure, but I think I'll stay aware from installers...

Website · Post by **Z-Man** » Sun Mar 06, 2005 11:03 pm

n54 wrote:wow "release quickly and release often" indeed

Um, yeah. I kind of got carried away. I must remember next time not to fix old bugs in the release phase, only new ones. To my defense, most of the bugs were new to me, at least.

great work, so far rc5 was the best imo - i'll get rc8 now

Thanks for the info!

Tank Program wrote: Build binaries? Sure, but I think I'll stay aware from installers...

I admit the installer packaging process could be more straightforward, but the NSIS scripts are really foolproof. They are a real improvement over the ClickTeam installer process ( They should be called ClickClickClickTousandClicksTeam.... ). BTW, Thanks, klax! For the next release, we'll certainly have a fully automated process.

ishAdmin · Post by **ishAdmin** » Mon Mar 07, 2005 1:29 am

rc8 was less sticky than 5 (i think, I'm getting confused as to which is which)

all of them have been pretty decent however. performance has improved in the area of cornering and tunneling. quite noticably.

I didn't notice much difference in play between them otherwise. this is on a poor and tired old windows 98.

klax · Post by **klax** » Mon Mar 07, 2005 6:32 pm

A bit late but here are the debs for Ubuntu Warty 4.10 i386 (or debian) of RC8:

n54 · Post by **n54** » Mon Mar 07, 2005 6:43 pm

ishAdmin wrote:rc8 was less sticky than 5 (i think, I'm getting confused as to which is which)

all of them have been pretty decent however. performance has improved in the area of cornering and tunneling. quite noticably.

I didn't notice much difference in play between them otherwise. this is on a poor and tired old windows 98.

I've had very little stickiness in rc8 playing at swampland, i'm using the term stickiness for when you stick and your movement doesn't make a difference. none at all while grinding.

however there is "stickiness"-ness (sorry for raping the english language) in corners or moving into a wall which is something slightly different i think: it enables you to use or "see" your rubber/gives your more time to change direction and as a result one can perform very neat highspeed manouvers in a series of very narrow tunnels and corners at highrubber servers like swampys. right now it can be an unfair advantage (but it easily backfires too) but that will disappear if it's available to all (after the release).

of course i personally think this is a feature and improvement of the game even if it might be unintentional

*hears z-man dragging his hair "not another feature fix!"*

Website · Post by **Z-Man** » Mon Mar 07, 2005 7:15 pm

n54 wrote: of course i personally think this is a feature and improvement of the game even if it might be unintentional *hears z-man dragging his hair "not another feature fix!"*

No, actually this is intentional. My new philosophy is that the official client should "cheat" as much as possible so there's no point in creating an extra-cheating hacked client

So, after I fixed the client rubber bug ( you certainly noticed that you often tunneled through walls even though you had plenty of rubber left ), I disabled client rubber accounting for the game mechanics completely. Unless the server tells you you're dead, your client should not give up hope.
The new server code is permissive, too. When you now turn slightly after you passed a wall ( because you're using an old client or someone just cut off your path ) and the server notices that and you turn early enough before the server rubber runs out, you'll still make the turn in time. Of course, there'll be lag sliding and such because the server is unable to execute your commands exactly, but you're still alive.

n54 · Post by **n54** » Mon Mar 07, 2005 7:24 pm

that's *insert any superlative of your own choosing here*

the dev team rocks!

Post by **Tank Program** » Mon Mar 07, 2005 7:24 pm

z-man wrote:I disabled client rubber accounting for the game mechanics completely.

Is this gonna mess up the HUD?

ishAdmin · Post by **ishAdmin** » Mon Mar 07, 2005 9:01 pm

I like how I can keep going, and know exactly where I might have died earlier if I didn't manage to survive. Before, I would often be unsure as to where or how I died. The way it is now is much more clear. It warps less, the explosion placement is accurate. My client not guessing is a very good thing.

z-man wrote:
I disabled client rubber accounting for the game mechanics completely.
Tank: Is this gonna mess up the HUD?

I have noticed something on some servers. The rubber seems to inexplicably run out suddenly. I think I have more, and then boom. It didn't happen on swampland much, but it has a rubber of 10. It seemed to happen with lower rubber servers. I would notice it most often when doing a 180 against a trail to go back down inside a tunnel. I think I have plenty of rubber for it, but I don't.

I don't consider this a big deal however. I think it's a matter of getting used to. I've never keenly watched the rubber meter, my observations are based on feel. By 'feel', or 'experience' I will know how much rubber I have left. I may have looked at the rubber going into a corner or tunnel, but I certainly don't look at it during the turns. After a period of adjustment, I will probably be able to 'feel' it better.