Going through rim walls/rips --> Lag? Bug?

[Lucifer]

Here's screenshots of the rendering bug I mentioned, but there was no arena rip at the time. Or at least, nobody said so when I asked. (people don't realize I don't get ripped locally I guess)

edit: Hey tank, can we get the bmp extension allowed for uploads? Or rather, if there's no particular reason it's not allowed, can we? (So you can discard thsi request if you specifically disallowed them for some reason)

[n54]

n54:
Are you using an older client ( not CVS )? It may in fact be that the new server side code will make ripping older clients easier. But i'm afraid I could only stop this if I disabled wall grinds ( you would need a min distance to walls to survive ).

Yes, i was using the 2.7 client and i suspect Nemo also was (?)

You consider that a feature? Rolling Eyes Oh dear. Another BUG_ variable I'll have to introduce.

I'm sorry but i can't stop loving those bugs that turn into features and i attribute it to your creative unconscious

And i kinda agree with lucifer &ish that as long as rips don't kill the server it's nothing to get upset about but we probably need specialized servers for those who like to go on ripping marathons because it can be very distracting.

just as we cannot terminate all lag i don't think we should aim to terminate all bouncing and similar effects. yes it can make the game unpredictable, and it sure gives people the opportunity to vent their emotions lol but i really think it's a BIG part of what makes this game so incredibly addictive. my point: don't eliminate it all just to eliminate it, eliminate the things that really counts to everybody (like rendering bugs like invisitails and tailforks)

"perfect" games are for wuzzies j/k

the imperfections of armagetron makes it more perfect than the average game imho

[Z-Man]

I realize if you did all that you could just as easily run your own test server...

That's actually a good idea. I'll run one on my old laptop with the relatively low bandwidth internet connection in gdb; the extra latency should be good for testing
I'll rather not ssh to your server and build it. I'm a lousy server admin, so I'll be afraid to break things all the time.

[Z-Man]

Lucifer, about those rendering bugs:
They sure look like sparks to me ( you turned off the floor rendering yourself, right? ). If you were using the same client that randomly bounced your rubber around, then those two are probably related. Spark positions and rubber detection both use the eSensor feature to detect walls ( that in turn uses game object movement code ), and that mechanism was broken at some time.

[Tank Program]

bitmap files tend to be huge, so for our dialup users like iceman... they've been disallowed

or they were disallowed by default and ive just left it that way, because bitmaps are huge

[Lucifer]

Lucifer, about those rendering bugs:
They sure look like sparks to me ( you turned off the floor rendering yourself, right? ). If you were using the same client that randomly bounced your rubber around, then those two are probably related. Spark positions and rubber detection both use the eSensor feature to detect walls ( that in turn uses game object movement code ), and that mechanism was broken at some time.

The screenshots may not show it, but those sparks are outside the grid, on the *other* side of the grid wall from us. The reason I thought they were associated with a rip is because whenever Nemo ripped my server the other night, those dots were there.

Yes, I turned off floor rendering myself (with 2+ people on my server, floor rendering kills my fps too much).

And yes, those dots look like sparks. The catch is that they're outside the grid. It's the same client that I was having those random rubber bounces too.

[dlh]

I'm using the 2.6 client, but the same stuff happens to me in 2.7.

And the rips that happen on BIH are hard to avoid. Even when I am just playing, I'll go to grind the outer rim and I'll rip. I can't avoid it. I'll try to compile the cvs client sometime this week, but I've had trouble compiling 2.6 and 2.7 for OS X. Theres always some unused variable or something that makes the build fail

[Lucifer]

I'm using the 2.6 client, but the same stuff happens to me in 2.7.

And the rips that happen on BIH are hard to avoid. Even when I am just playing, I'll go to grind the outer rim and I'll rip. I can't avoid it. I'll try to compile the cvs client sometime this week, but I've had trouble compiling 2.6 and 2.7 for OS X. Theres always some unused variable or something that makes the build fail

Is there somebody maintaining an OS X build system? If not, would you be willing to do it, nemo? It would be nice to have more solid OS X support...

[Z-Man]

Yes, and even if you don't want to maintain the build, you should report compile errors. The rest of us does not see them, you know...
Hmm. we could just try to pretend to be building for mac, that could at least catch trivial errors like those missing variables.

[Tank Program]

We could compile in PearPC if one of us has a copy of OS X lying about. In fact, that's how I did the release build for OS X. Compiled in an emulator. In like 7 hours...

[Jonathan]

I'm using the 2.6 client, but the same stuff happens to me in 2.7.

And the rips that happen on BIH are hard to avoid. Even when I am just playing, I'll go to grind the outer rim and I'll rip. I can't avoid it. I'll try to compile the cvs client sometime this week, but I've had trouble compiling 2.6 and 2.7 for OS X. Theres always some unused variable or something that makes the build fail

I haven't tried CVS (yes, not even CVS itself), but this is what I did:
- Add some missing files to the pbproj.
- Put SDL, SDL_image and SDL_mixer in ~/Library/Frameworks/ and make sure the paths in the pbproj are OK.
- Maybe some other things...

Maybe I can help [with the OS X version], but I'm far from an experienced programmer. I also still don't know what's actually wrong (when running, compiling is fine) if I build using the optimizer (at least the "for size" setting, I haven't tried all).

[Z-Man]

We could compile in PearPC if one of us has a copy of OS X lying about. In fact, that's how I did the release build for OS X. Compiled in an emulator. In like 7 hours...

You're crazy

Jonathan: If that was you crashing my CVS server, please do it agan! The debugger crashed while analysing the bug... I've got a pretty good idea what happened, though.
For the others: Jonathan managed to crash the topology police...
Edit: nevermind, I could reproduce it locally by simulating the old client behaviour.

[Z-Man]

This bug Jonathan uncovered is really interesting, but I won't describe it here until Lucifer updated Hell. It is a crash caused by a situation that should be impossible.
The fix is in src/engine/eGrid.cpp revision 1.17.

[Jonathan]

So this is why I suddenly lost the connection... I didn't even know that I crashed the thing.

Nice to help you anyway.

[dlh]

I'm using the 2.6 client, but the same stuff happens to me in 2.7.

And the rips that happen on BIH are hard to avoid. Even when I am just playing, I'll go to grind the outer rim and I'll rip. I can't avoid it. I'll try to compile the cvs client sometime this week, but I've had trouble compiling 2.6 and 2.7 for OS X. Theres always some unused variable or something that makes the build fail

I haven't tried CVS (yes, not even CVS itself), but this is what I did:
1) Add some missing files to the pbproj.
2) Put SDL, SDL_image and SDL_mixer in ~/Library/Frameworks/ and make sure the paths in the pbproj are OK.
3) Maybe some other things...

Maybe I can help, but I'm far from an experienced programmer. I also still don't know what's actually wrong (when running, compiling is fine) if I build using the optimizer (at least the "for size" setting, I haven't tried all).

What version of gcc are you using? I'm using 3.3. Ill try using an older version.

Edit: found those missing files causing those undefined symbols. Thanks Jonathan