Re: [AD] Allegro unable to deal with non ascii filenames

[Chris <chris.kcat@xxxxxxxxxx>]

On Saturday 27 May 2006 14:48, Elias Pschernig wrote:
> On Sat, 2006-05-27 at 22:40 +0200, Michal Molhanec wrote:
> >    Hello,
> > don't tell me that there is no way on Linux how to ask system which
> > system encoding is used and/or to convert string from/to UTF-8 to the
> > system encoding.
>
> Well, probably all libc functions will use the current locale setting.
> So if your locale is UTF8, filenames will be returned as UTF8. If it is
> latin1, filenames will be latin1.

Wouldn't it be possible to read this info from the LANG environment variable? 
For me, this is LANG="en_US.UTF-8" and passing UTF-8 data to the file 
functions seems to work just fine (filetest shows all the info properly, and 
ufile.c was changed to pass UTF-8 data instead of ASCII.

I'm not really sure how to handle this, though. Just creating a new function 
won't relly help because it would need to be called with a specific value 
between systems. Perhaps it could just use a new internal setting that's set 
when allegro_init is called, using the LANG variable. Then when accessing 
files, it'd convert to that setting. IIRC, the Mac port uses something like 
_al_open for some internal magic, so perhaps this could be extended to 
Windows to convert filenames to UTF-16 and use _wopen on NT/2k/XP, and on 
Unix/Win9x convert to the current system locale setting and call the plain 
open function.

Sounds a bit too much to do for 4.2.1 though, IMO.