[BRLTTY] What this means?

[Samuel Thibault]

Dave Mielke, le Tue 25 Aug 2009 12:03:09 -0400, a écrit :
> [quoted lines by Samuel Thibault on 2009/08/25 at 14:58 +0200]
> 
> >Where should it not work? iconv can convert it, and the xlate insertion
> >case should be able to insert the characters one after the other.
> 
> Part of it is our own deficiency in that convertWcharToChar() is only designed 
> for single-byte character sets. I think there may be another part, though. It's 
> not clear to me what their keyboards generate. Perhaps it's BIG5, but we should 
> probably verify this.

Since the mode is xlate, it should just be what applications expect,
which is just the current locale's charset.

> There may be another aspect as well. I believe they work in at least three 
> character sets (BIG5, GB2312, UTF-8) simultaneously.

That should never be simultaneously. Documents may be in different
encodings, but the very notion of a locale imposes just one character
set for applications' input/output, and applications just convert from
the document charset to the locale charset. Else for the screen it's
just guesswork to know how characters should be displayed.

> They do this by telling the screen driver which character sets to     
> check for via its charset= parameter.                                 

Yes, and that is never to be working properly. It happens that it
probably more or less works, because valid utf-8 will often not be
valid GB2312 and vice-versa and brltty happens to kind of automatically
switch, but between GB2312 and BIG5 that's not so clear, as shown by the
problem he raised on the list last month
http://mielke.cc/pipermail/brltty/2009-July/005944.html

Actually, the best charset that should be used for chinese locales is
probably just GB18030, which is a superset of GB2312 and includes all
unicode characters.

Samuel