On Thu, Jan 22, 2004 at 01:47:27AM +0400, David Tolpin wrote:
> Rich Salz:
> > Why not just do everything in UTF-8. All this wchar_t, "asdfasdf"L
> > support, etc., is going to complicate the API. Seems simpler to have
> > the basic API, and for those not working in UTF-8 (or its subset:), they
> > only have to remember a single consistent rule: call "convert()" first.
>
> That would be the best. I wanted to suggest that but thought it was stupid.
> At least I am not alone.
I was looking at expat.h to see what James did. Expat automagically
switches between UTF-8 and UTF-16, and that's a build time option.
What's the use case for an app to use both UTF-8 and UTF-32 concurrently
with genx? I've been using expat since approximately the beginning of
time, and I've only dealt with a single encoding (UTF-8). Having the
API unified behind a single XML_Char type -- UTF-8, UTF-16, UTF-128 or
whatever -- certainly helps.
So, are the genx*W API calls even necessary? Can all of this complexity
be hidden behind a single typedef like James' XML_Char? Actually, can
it be hidden behind that specific set of typedefs? Anyone who really,
desperately, needs to avoid the encoding genx uses will know what to do.
Z.
