Frikqcc problems?

[metlslime]

I'm using Frikqcc 2.7, which seems to be the latest. Basically, i have intermittent problems where if i compile the code and run a map (any map), the command will fail with a message like these:

"couldn't spawn server maps/sub_checkrefire.bsp"
"couldn't spawn server maps/ftos.bsp"
"couldn't spawn server maps/testplayerstart.bsp"

Making additional code changes and recompiling will usually fix it, but then it comes up again in the future.

On a hunch i put a call to runerror() in GotoNextMap (the only place in the source that calls "changelevel", and it produces this call stack:

GotoNextMap
ExitIntermission
ClientConnect

I put in a dprint to see the value of intermission_running when ClientConnect is called, and the value appears to be "-60"... and the value of "nextmap" appears to be "sub_checkrefire" -- so that explains the symptoms, but what would explain these garbage values?

Anyone see problems like this before or know what causes them? At the moment, i am back to things working again, but it would be nice not to be at the mercy of things magically breaking and fixing themseves.

[frag.machine]

Smells like globals area corruption, maybe due some compiler optimization.

[Spike]

make sure numpr_globals doesn't exceed 32k.
Not only do most engines not support that, but I remember frikqcc having a bug in that area, but that might have been version 2.6.

[metlslime]

make sure numpr_globals doesn't exceed 32k.
Not only do most engines not support that, but I remember frikqcc having a bug in that area, but that might have been version 2.6.

Thanks, i'll check that out.