wchar_t working on sparc-sun-solaris2.6

[Branko]

Benjamin Kosnik wrote:
> For right now, I'd like to become more strict about enabling wchar_t. I'm
> not quite convinced that replacing w* functions if they don't exist is
> really the best solution.
> 
> I'd rather get wchar_t working correctly on the platforms that that have
> the infrastructure to correcly support wchar_t, rather than doing a
> mediocre job everywhere. After than, ports can happen....

O.K., I agree with that. In the meantime, I'll see if TODO has
anything I could help with ...

> 
> Anyway.
> 
> > On another note: I see that most of the work currently being done on
> > libstdc++ is specific to glibc-based platforms (or at least those with
> > a libc that tracks the C standard closely enough). This will probably
> > cause problems on most commercial and older Unices, which usually=20
> > concentrate on backward compatibility. I didn't find any plans for
> > porting the library to other platforms on the web pages, and I have a
> > strong feeling that, the way things are going now, the library either=20
> > won't be ready in time for GCC 3.0, or it will be useful only on GNU
> > systems -- Linux and Hurd -- as I don't believe any others use glibc.
> 
> The idea is to use a modern, synched libio for these platforms. This is
> the current approach for libstdc++-v2, and it's hoped that v-3 can offer
> the same coverage.

It would be nice if something like that could be available on all
platforms, but I guess it's impossible without support from the C
library (at least if you want reasonable performance).

> > I do hope my perception is wrong... but I'm a bit worried. I'm prepared
> > to help with the implementation on Solaris 2.6 and HP-UX 10.20, within
> > the limits of my free time.

    Brane

-- 
Branko Čibej                 <branko.cibej@hermes.si>
HERMES SoftLab, Litijska 51, 1000 Ljubljana, Slovenia
voice: (+386 1) 586 53 49     fax: (+386 1) 586 52 70