[MLton-user] Re: MLton signal handling broken on Windows (was: CML and foreign code)

[Wesley W. Terpstra]

On Wed, Oct 5, 2011 at 6:10 AM, Ivan Tomac <ivan.tomac at gmail.com> wrote:
> On Wed, Oct 5, 2011 at 4:29 AM, Wesley W. Terpstra <wesley at terpstra.ca> wrote:
>> Do you have an unusual version of Windows that I might never
>> have used MLton on?
>
> I can confirm that I can't get the problem to happen on a computer
> running Windows 7 32-bit.
>
> I have tried the code on 2 machines running Windows Vista 32 SP2 and
> Windows Vista 64 SP2 and I get crashes on both. So I think it has
> something to do with Vista, or at the very least Vista SP2.

Ok. I never had Vista myself, so I never personally tested MLton
there. I do know other people have reported using it on Vista, but I
don't know how thoroughly they've tested it. I have no idea where I'd
get my hands on a copy of Vista either.

Was SP2 for Vista only recently released?

> Removing
> SetThreadPriority seems to make the problem disappear but I'm worried
> there may be a more serious problem there - maybe the problem doesn't
> actually disappear but becomes less frequent?

That would be my guess.

> I am not sure if any processes could be interfering. As for virus
> scanners, I am running Microsoft Security Essentials anti-virus on all
> my Windows boxes.

Ok, I use the same virus scanner without problems.

> I don't think it's bad memory as I can replicate the problem on 2
> different computers and I haven't noticed any problems running
> processor and memory intensive applications on them.

Ok. That seems quite conclusive.

>  Fault Module Name:    ntdll.dll
>  Fault Module Name:    ntdll.dll

Interesting. It's faulting inside a system DLL.

Would you be able to checkout the source code for MLton and run the
regression suite?
In the msys shell, just enter the MLton source top-level directory and
run ./bin/regression