On Tue, Feb 1, 2011 at 1:57 AM, Igor Stasenko <siguctua at gmail.com> wrote:

> Yes, this is one thing, which i discovered during NativeBoost development.
>
> I made the patch which fixes it
>
> http://code.google.com/p/nativeboost/downloads/detail?name=001-perform-framesize.cs
>
> Note that it deals with #perform:withArguments:
> but not with #perform:withArguments:inSuperclass:
>
> means that if you use the #perform:withArguments:inSuperclass: , it
> also may fail, depending on call site, which may not fit all arguments
> into
> context stack. See senders of it.
>
> The main reason why it works on Cog i think, that Cog maps contexts to
> stack.. so contexts don't have  a hardcoded limits of their stack
> sizes.
> (but that juts guess , i didn't checked it)
>

That's right.  This applies to both the JIT and the StackInterpreter.


>
> On 1 February 2011 00:09, Nicolas Cellier
> <nicolas.cellier.aka.nice at gmail.com> wrote:
> > Here is an interesting bug solved in Pharo affecting Squeak too with
> > regular (non cog) vm
> > http://code.google.com/p/pharo/issues/detail?id=3468
> >
> > A patched CompiledMethod>>needsFrameSize: is in Pharo.
> >
> >
>
>
>
> --
> Best regards,
> Igor Stasenko AKA sig.
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.squeakfoundation.org/pipermail/squeak-dev/attachments/20110201/7b0a9e95/attachment.htm