[Uni-verse] possible maxbug

[Martin]

> 
> Only now, been out of town for a while (Amsterdam!). This thread really
> sounds like it should have been on the verse-dev list, but I guess you
> can't have everything.
> 
> Martin's experiences with crashes on "fast" machines are really weird,
> I don't have any relevant experience I think. The machine I'm using
> here at work is Intel, 3 GHz Pentium 4 with hyperthreading enabled. I
> haven't seen any of these mysterous crashes. Not sure what to think
> about the timer wrapping theory, either. The performance counters are
> larger than 32 bits, which should help. Of course, I'm not a Windows
> person, so I might miss stuff.

Dual core, Win32, Intel.


> I don't quite get this part, either. I know the code still contains
> stuff like that, although many have been cleaned out by me during the
> project. Can't see why it should halt the application though, but per-
> haps that's the debugger being very sensitive. Again, I don't have a
> lot of relevant experience with Windows tools in this regard.

I have never seen anything like it; I mean such a warning throws such an
exception/halt.
The debugger is not very sensitive.
In this project I compile 10 other libraries with total 250 000 lines of
c/c++ code used with the verse code, all with the same debug/compile/link
settings, only verse halts.
When I have time I'll check into your code a bit and try to verify.

Anyway, I have had a lot of problems with linking the verse lib on win32.
I suspected that you made some kind of static linking to libcmt (standard c
library)? Pragma link?
It is also weird, since we have to aggregate the code to be able to make it
link to a multithreaded standard c/c++ library, we will get conflicts
otherwise.

Maybe these two problems have something in common... 

/martin