[Vorbis-dev] Re: vorbis-tools 1.2.0 Release Candidate

[ogg.k.ogg.k at googlemail.com]

> You are being a bit vague here. Are you talking of recent work or
> code that has been there forever?

I don't know when that code went in, I'm talking about the code I found
in theora.c, that was changed by my patch. It may be recent or old, I
don't know. My point was that the fix I did yields more simple code.

> > ogginfo works on my (32 bit) machine for theora comments.
>
> Oh, that's accidental, I assure you. You see, ogginfo was
> specifically optimized to work on 64bit SH4 CPUs. It must have been a
> recent patch that made it work on those pesky old 32bit machines you
> speak of.

Good, good :) I was just saying that I can't test my patch on a 64 bit arch,
but I at least tested on my 32 bit one :) This being relevant because the
patch attempts to fix a problem that would show on 64 bit archs.

> Seriously now, was it not working for you until you changed "len" to
> long? Or was that change to fix a warning?

It was working fine, but I reckon it wouldn't have worked on a machine
where sizeof(long)>sizeof(int). But I can't test, so I'm not 100% sure.
Unless I'm missing something, writing 64 bits into a 32 bit space will just
smash whatever's in the "other" 32 bits, and it's on the stack, so danger
is around the corner.

I apologize in advance if I'm having the wrong end of the stick here.
Maybe it's not a problem and I'm just very confused.