Problem solved...

ANOQ of the Sun
Mon, 05 Jun 2000 00:48:36 +0200

"Stephen Weeks" wrote:
> > I have just seen the binary make a segmentation
> > fault on an unhandled exception in the ML code.
> > This should not happen - right?
> Right.  There is supposed to be a default exception handler installed
> that prints out an "unhandled exception" message.  If this error is
> repeatable, could you please send the SML code.

I got that error "unhandled exception" message first -
then the seg-fault. But I only saw it once when I had
the debug messages in the io-code - so I can't reproduce it.

> As to the rest of our recent email exchanges, my impression is that
> this is the only remaining MLton bug/question.  Please remind
> me if there are others.  Thanks.

Well, right now I have the GC-problem under Windows again -
and not under Linux.

I'm sending you (Stephen) an exe-file and a Lightwave object
file which gives the problem. The program is actually going
to be commercial - but since it does not work I don't mind
sending you the .exe :)

Execute it like this:

CeX3DConverter @MLton gc-messages gc-summary -- Triceratops.lwo

I've added a message at every VirtualAlloc and VirtualFree.
It seems to fail after the messages:
Shrinking stack to size 262144
unable to VirtualFree memory at 5ef0000

And by the way - the "unable to VirtualFree" address is hex,
the other virtual alloc/free are decimal.

Do you have any ideas about what this is?