Re: [xml] setting the default charset ?

[Cyrille Chepelov <chepelov calixo net>]

Le sam, jui 28, 2001, à 10:14:38 -0400, Daniel Veillard a écrit:
> On Sat, Jul 28, 2001 at 11:23:23AM +0800, William M. Brack wrote:
> > xmlDetectCharEncoding looks at those four characters, notices that they are
> > "<!xm", and
> > then sets the encoding for the context to be UTF8 (somehow this seems
> > related to the
> > classic expression of the american Henry Ford, who once declared "You can
> > have any
> > colour you like, as long as it's black") (no offense intended).
>
>   It's the normal process as described in the XML specification
> there is an appendix about charset detection, though not normative
> it's a good idea to follow it. If the application didn't tell 
> what encoding is the entity, it's assumed to be UTF8 or UTF16
> unless overriden by the encoding decalration in the XML decl.
> Not a bug, a feature !

Actually, more probably a bug. Let's see :

        1) I  do a xmlSwitchEncoding(something like KOI8-R);    
        2) then I start parsing.
        3) xmlDetectCharEncoding sees <?xm
        4) it decides <?xm is UTF-8 encoding, so it switches to UTF-8, since
           UTF-8 is not NONE.   
        5) my setting of KOI8-R is forgotten.

<disclaimer>I might read that code differently from gcc.</disclaimer> But I
think William spotted the problem.

What could perhaps be done would be to apply the encoding detected in 4) only if we
detect something else than UTF-8 or we don't have a previous encoding. If we
have a non-default, and we detect a 7 or 8-bit file, then we use the
(previously) user-supplied charset and don't use what we detected. At least
until we find an encoding="..." attribute.

        -- Cyrille

-- 
Grumpf.