Re: [Pnet-developers] Porting work remaining question

[Peter Colson]

On 05/10/2004, at 4:25 PM, Gopal V wrote:

> Hi,
>
> > It seems that once I head out of local object
> > territory into the System
> > area via WriteLine that's when the problems begin.
> >
> > Any thoughts on the matter?
>
> It's the CodePage junk that's choking your box.
> Like I said before HelloWorld's no longer a
> trivial example in the Unicode world :)

You say there's a lot of codepage stuff going on, but
what I need to know are some reasons this might fail?

I see similar CVM dump contents under OS X up to
the point of failure but it continues to produce >3,000
lines of CVM dump. Why would this be different - is pnet
under Mac OS X running (or not) as Unicode - I thought
it might be?

Are there any steps to be taken to get a Unicode-specific
system working? I note the support/mkcase, mkcategory,
mknumber programs that create values from UnicodeData.txt
(a file that I didn't find in DotGNU-0.6.8). Is there anything
that I need to do with these programs?

I'm very much at a loss as to what I can do at this point...

Symbolically debugging into/through cvm.c/cvmc.c is nigh
on impossible with the excessive use of includes/macros
in the interpreter switch.

I'm concerned that the different pointer sizes and pointer size
being different to int/long may be an issue, but don't know
how to prove this one way or the other.

> I'm still not sure that the CVM code generation
> is right (obviously). The cvmc_setup for the
> case where INT32 is not native , has a
> CVM_ADJUST(-CVM_WORDS_PER_LONG) , which I find
> very strange. Any comments, Rhys ?.

Removing CVM_ADJUST(-CVM_WORDS_PER_LONG) in the
only area I saw it in cvmc_setup near a mload/l2i had no effect...


Regards,
Peter Colson.