Possibility of an extra device?

Nick Burrett nick at dsvr.net
Wed Feb 5 01:31:47 PST 2003


John Tytgat wrote:
> In message <3E3FC80F.8000408 at dsvr.net>
>           Nick Burrett <nick at dsvr.net> wrote:
> 
> 
>>I have a steadily increasing patch to fix the libstdc++ stuff, but it is 
>>slow going as I'm having to implement missing wide-character functions.
> 
> 
> I was thinking to have a go at this myself but now I see that you took
> the challenge... ;-) Seriously, my current feeling about the wide-char
> support was we need an iconv module which could be used by UnixLib.
> As the iconv routines potentionally need to make use of large memory
> blocks and these better get shared between the running UnixLib based
> programs.
> 
> Makes sense ?

Yes. I agree and did have a unixlib tree with the 'iconv' stuff in, 
though I accidently deleted it.

I was thinking of porting 'libiconv'[1] but I'm quite fearsome of all 
the multibyte encoding stuff.

The wchar implementation that I'm doing is written to support 
MB_CHAR_LEN of 1, though I'm trying to ensure that the work required to 
support a different MB_CHAR_LEN is restricted to only a small set of 
low-level functions.

I'm basing the work on the libc from FreeBSD-current CVS tree because I 
think deriving from GNU libc is going to overcomplicate the library.

At least if I get the basic wchar functions complete, then we have a 
starting point to further extend if we feel it is necessary.

[1] http://www.gnu.org/sotware/libiconv/.


-- 
Nick Burrett
Network Engineer, Designer Servers Ltd.   http://www.dsvr.co.uk




More information about the gcc mailing list