Enrico Forestieri wrote:
On Wed, Oct 31, 2007 at 06:14:35PM +0100, Abdelrazak Younes wrote:
Jean-Marc Lasgouttes wrote:
Enrico Forestieri <[EMAIL PROTECTED]> writes:
Every character you output to a docstream must be ucs4 encoded.
If its code point can be contained in an unsigned char, fine,
otherwise you need a larger type.
Hmm. I think this is a recipe for hard to find bugs.
That's what I am trying to tell to Enrico but he doesn't seem to hear
me :-)
The recipe for bugs is not considering the possibility to be
able to output a unsigned char to a docstream. And this is
not hypothetical, as we already have a bug due to this fact.
Actually I don't think your patch is better than Jurgen's patch and I
don't understand why Jurgen's patch would not work under platform were
wchar_t == 2. Jurgen's patch seems correct to me and if it fails on your
platform it must be because of another reason.
As for adding the possibility to output an unsigned char to the
docstream I think this buy us absolutely nothing and that it is not even
worth supporting.
Abdel.
