Re: [xml] encoding practice?

[Hannu Krosing <hannu tm ee>]

On Fri, 2002-12-13 at 12:52, Daniel Veillard wrote:
> On Fri, Dec 13, 2002 at 07:29:17AM -0500, Derek Holden wrote:
> > I am wondering what common practice is for reading and writing XML 
> > internally. Specifically, xmlDumpMemory writing utf8 and xmlParseMemory 
> > reading iso-8859. Is it better to handle the decoding on the dump end or 
> > encoding on the parsing end? If I'm approaching this incorrectly or there 
> > is an equivalent utf8 xmlParse routine I'd appreciate hearing it. Thanks.
>
>   I would say keep everything UTF8, UTF8 and UTF16 are the only encodings
> that any XML parser MUST support. And I would advise against UTF16 in general
> because it forces extra conversion in most processing tools and in general
> wastes spaces with useless zeroes...

AFAIK this is only true for latin-based scripts. For most far-east
scripts UTF-16 actually saves space by spending only 2 bytes per char
instead of 3.

BTW, is there any support for using UTF-16 internally in libxml2/libxslt
?

> Daniel
-- 
Hannu Krosing <hannu tm ee>