New Armagetron - Locks Up

[Spud Patrol!]

Got the new one, didn't install it and then upgraded my PC to have XP Pro. I suspect something with the firewall is afoot, but before I start setting up my router to do other than 'automatic firewalling' (Solwise 715x), ie individual ports settings, I thought I should ask here.

Under Win 2000 I didn't need to do anything but install the game and run it. No problems in single, lan or internet game mode. Under XP it's a different story.

Armagetron Advanced gets to the point where it says "47 Servers responding" and then it locks tighter than a bull's butt at fly-time. Obviously it gets out of the confines of the firewall enough to find out how many servers there are to play on. What's wrong then? Any command-line stuff or diag codes I can go for?

On a peripheral note why isn't there a better way to set up cycle/trail colors yet? Thanks in advance.

[Z-Man]

Sadly, errors of this kind are almost impossible to track down even with physical access to a machine. Most of the time, they vanish in debug mode. So, although there are similar reports for all OSes, we're still pretty much clueless about the cause. To work around the problem, you may try
- different screen modes
- disabling sound
or download the OpenGL call logging client from here: http://guru3.sytes.net/viewtopic.php?t= ... c&start=30
and follow the instructions for Seeker. He reported a different kind of error in a similar situation. Note that the log file grows big very quickly.

Feedback on how things went will be greatly appreciated.

On a peripheral note why isn't there a better way to set up cycle/trail colors yet? Thanks in advance.

Because you're not supposed to have a cycle color different from your trail color in the first place Strictly speaking, the way you do it now is a hack or cheat. A bug exploit. It'll stop being fun once it's official, IMHO.

[Spud Patrol!]

Thanks in advance..

Sadly, errors of this kind are almost impossible to track down even with physical access to a machine. Most of the time, they vanish in debug mode. So, although there are similar reports for all OSes, we're still pretty much clueless about the cause. To work around the problem, you may try
- different screen modes
- disabling sound

Did that. I should have mentioned that if one plays a single-player game it works fine. This is Internet play-based.

or download the OpenGL call logging client from here: http://guru3.sytes.net/viewtopic.php?t= ... c&start=30
and follow the instructions for Seeker. He reported a different kind of error in a similar situation. Note that the log file grows big very quickly.

I've got enough disk space to do the above.

Feedback on how things went will be greatly appreciated.

On a peripheral note why isn't there a better way to set up cycle/trail colors yet? Thanks in advance.

Because you're not supposed to have a cycle color different from your trail color in the first place Strictly speaking, the way you do it now is a hack or cheat. A bug exploit. It'll stop being fun once it's official, IMHO.[/quote]

I'd hardly call color-setting a cheat, though. My main thought on this had to do with the semi-florescent colors one could obtain with the numbers greater than 15, regardless of the trail.

[Spud Patrol!]

Sadly, errors of this kind are almost impossible to track down even with physical access to a machine. Most of the time, they vanish in debug mode. So, although there are similar reports for all OSes, we're still pretty much clueless about the cause. To work around the problem, you may try
- different screen modes
- disabling sound
or download the OpenGL call logging client from here: http://guru3.sytes.net/viewtopic.php?t= ... c&start=30
and follow the instructions for Seeker. He reported a different kind of error in a similar situation. Note that the log file grows big very quickly.

Feedback on how things went will be greatly appreciated.

Completely removed all versions of our favorite game, no folders no nothing. Checked registry, nothing there. Clean installation of the current version, and the call logging client url'ed above. Single-user works fine but occasionally crashes with graphics error taking up the top half, and showing the desktop scrunched in the lower half of screen. Network play does this 100% of the time, but now crashes back to Desktop after quickly showing an error message I couldn't catch, only occasionally doing the total lockup with halved screen, requiring hard reboot.

The Stdout file is attached. Thanks for your help!

[Spud Patrol!]

Did the error messages tell you anything?

I fiddled with the settings for OpenGL and found that the stdout file was smaller by 2k. Still locks up the same way though.

I really miss playing Armagetron. Help!

[Jonathan]

First, they're not error messages. Second, it doesn't look like the problem is related to OpenGL.

On a peripheral note why isn't there a better way to set up cycle/trail colors yet? Thanks in advance.

Because you're not supposed to have a cycle color different from your trail color in the first place Strictly speaking, the way you do it now is a hack or cheat. A bug exploit. It'll stop being fun once it's official, IMHO.

I agree.

[dlh]

Missing frommaster.srv file maybe? I have no clue where'd be on windows, though.

[Z-Man]

Spud: No, sorry, I'm still completely clueless. The last message says it will swap the GL buffers, there can't possibly be anything wrong with that. I'll have to check the possibility of the real crash message getting swallowed because it's still in the file buffer and not yet written to disk.

You don't overclock anything or use beta graphics drivers so we have something to blame it on?

Nemo: the frommaster.srv file is generated automatically, but only after the first connection to the master server completes successfully.

[Spud Patrol!]

You don't overclock anything or use beta graphics drivers so we have something to blame it on?

Nemo: the frommaster.srv file is generated automatically, but only after the first connection to the master server completes successfully.

No overclocking, no beta graphics drivers - though I admit that the ATI All-in-Wonder 128 Pro in this machine is going to go the way of all hardware in the next two months; it's capable to 4x AGP, and is seen as that by the BIOS... because of this however the system runs at 1.9GHz instead of 2.8 - something I think will be resolved when the card is replaced (most likely with a Radeon 9800, the one with the really good heat sink instead of a fan).

I at first suspected Windows XP but remembered that the former version of Armagetron .26 didn't have this problem. When I upgraded to the Advanced .27 this began happening. I even removed the package and reinstalled, to no solution.

If there is a problem graphically, wouldn't it show up in ANY kind of game, as opposed to a Network or Internet game? Thanks.

[Z-Man]

Although this underclocking is certainly strange, it's not anything we can blame this error on.

If there is a problem graphically, wouldn't it show up in ANY kind of game, as opposed to a Network or Internet game? Thanks.

At first sight, yes. But the error description you gave sounds very graphical in nature; I'd expect a networking error to manifest itself in a regular program abort (access violations are caught by SDL and handled gracefully). Networking activity changes the timing of the application and its OpenGL calls; real hardware or driver errors may be triggered by these subtle changes.
But of course, enough fishy things go on in the low level networking systems; under certain circumstances (that may very well happen in the internet server browser), invalid sockets may be passed to the operating system. That part of the code was not changed from 0.2.6.0 and to 0.2.7.1, however, but it's the only clue I've got. I only fixed it up recently in CVS. My guess is that Luke will want to have a prerelease binary with map support built soon, that will contain these fixes.

[Spud Patrol!]

Although this underclocking is certainly strange, it's not anything we can blame this error on.

If there is a problem graphically, wouldn't it show up in ANY kind of game, as opposed to a Network or Internet game? Thanks.

At first sight, yes. But the error description you gave sounds very graphical in nature; I'd expect a networking error to manifest itself in a regular program abort (access violations are caught by SDL and handled gracefully). Networking activity changes the timing of the application and its OpenGL calls; real hardware or driver errors may be triggered by these subtle changes.
But of course, enough fishy things go on in the low level networking systems; under certain circumstances (that may very well happen in the internet server browser), invalid sockets may be passed to the operating system. That part of the code was not changed from 0.2.6.0 and to 0.2.7.1, however, but it's the only clue I've got. I only fixed it up recently in CVS. My guess is that Luke will want to have a prerelease binary with map support built soon, that will contain these fixes.

Of course the AGP AIW Pro 128 is the reason why the machine's not hitting speeds above 2GHz - but it was after the upgrade that this problem began.

I'm beginning to wonder if something re: OpenGL might give a clue. What if I cleared and re-genned another stdout after changing my opengl settings somehow?

[Z-Man]

You can try, but I don't see much chance another run would give more information. I'll have to build a new version were all commands are guaranteed to be logged (nothing gets lost on the crash) first, probably.

[Spud Patrol!]

You can try, but I don't see much chance another run would give more information. I'll have to build a new version were all commands are guaranteed to be logged (nothing gets lost on the crash) first, probably.

Alas! I've tried Our Favorite Game with the 3D stuff partially-disabled, though I didn't keep notes (it was late), and at the point of 16-color it started not locking at that same point. Looked awful but hey!

Well, as a process of General Upgrades Before Failure, I'm replacing the AGP card in this machine with a stop-gap low-cost solution: a Gecube Radeon 9550 AGP 8x 256MB card via ebay. We'll see what this brings to this thread.

As an aside I'm eyeing up the 128MB/256MB Radeon 9800 series, mainly for the fan-less heat sinks. Any experience with these? No, I don't have £400 to spend on a pair of cards. Thanks.