> Obviously, to have gotten the assembly routine, I must have been debugging
:)
Not necessarily, some development tools modify the Windows crash message to
include it.
> These numbers are helpful if you have a debug version of the engine and
the
> source, like say a valve employee, as you can simply go to the memory
location and
> see what the code is, it will be the same location for any mod. While this
may not
> give the exact problem, it would be useful in determining what kind of
problem it
> is. Like with messaging or spawning etc.
That's assuming they're using an unmodified, stock version of the last
patch's build for development purposes, which is far from certain.
> Just to add to my paranoia the crashing has stopped ... I started
debugging the
> messages, which involves putting a wrapper around the definition of
REG_USER_MSG
> to get it to print out the message name/number pairs, and when I compiled
it no
> longer crashed.
>
> Yes I did do try a full recompile of both client and server before I
posted here,
> in fact I did a clean on both, then a full compile. No I didn't make any
other
> changes.
>
> Wonder if I fully patched VC++ since I last reinstalled windows ... I hate
it when
> things fix themselves, you never know when they will decide to unfix
themselves
> again :{
VC++ is notorious for this type of compiler-generated heisenbug; I recently
ran into one while editing pm_shared.c. The addition of an if statement,
simply reading a variable, with an empty conditional statement made the
thing work somehow. There must be some kind of quirk in the compiler itself
that causes these things that hasn't been patched by M$.
Good luck in not having the thing return. :)
----- Original Message -----
From: "Jeff "Codiac" Fearn" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Thursday, February 20, 2003 12:40 AM
Subject: Re: [hlcoders] Engine crashing on map change
> > Try debugging it in Visual C++ and checking the call stack to see where
the
> > problem is coming from. Those numbers are really unhelpful in locating
the
> > source or nature of the error unless you've decompiled HL and memorized
all
> > the compiler-generated assembly instructions in it. You could use some
other
> > debugging program if you want, VC++ is simply more efficient if you're
using
> > it to compile the code in the first place. The problem is obviously more
> > difficult to solve if the call stack doesn't help, but it's a good place
to
> > start any bug-squashing run.
>
>
> I should however have mentioned that the stack only contained a reference
to
> CSTRIKE!
>
>
> > > > I get a crash in cstrike.exe every time a level changes, CSTRIKE!
> > 01da7c55().
> > > > 01DA7C55 call dword ptr [edx+4]
>
>
> Jeff.
>
> _______________________________________________
> To unsubscribe, edit your list preferences, or view the list archives,
please visit:
> http://list.valvesoftware.com/mailman/listinfo/hlcoders
>
>
>
_______________________________________________
To unsubscribe, edit your list preferences, or view the list archives, please visit:
http://list.valvesoftware.com/mailman/listinfo/hlcoders
