Re: [xslt] output escaping (again!)

[Daniel Veillard <veillard redhat com>]

On Mon, Mar 24, 2003 at 09:15:35AM -0500, Bruce Miller wrote:
> Daniel Veillard wrote:
> > On Sun, Mar 23, 2003 at 06:07:34PM -0500, Bruce Miller wrote:
> >>Here's a simpler example to get to the point:
> >>  (4) data:text/plain,Hello
> >>  (5) data:text/plain%2CHello
> >>
> >>Case (4) is valid, but libxml won't let me create it.
> > 
> > 
> >   As I say and repeat, it's fixed in CVS.
> 
> Yes, the particular case of "," is -- and I appreciate that.
> But not the general case, which would be solved by _not_
> escaping _any_ non-ascii.
> 
> I sense I'm annoying you by continuing, but I'm just asking 
> for clarification:
>   Why do you feel that libxml should automatically
> escape any of the reserved characters at all?

  To avoid breakage and compatibility problems with things like
filename with spaces used as href in XSLT.

> [  It would seem that given str:encode-uri (in libxslt, and
> something similar in libxml), the programmer (stylesheet 
> designer,...) can do the encoding exactly where (and only
> where) needed.]

  And the 'where needed' from a programmer point of view will
be never since most of the tools completely dropped the notion
of URI conformances and take any string without asking.
  The answer is *cleanup* , it may break occasionally. Like
the fact that the URI serialization do the UTF-8 -> %XX transcoding
of URI at saving time even if some people complains that they
would like it in the document encoding.

Daniel

-- 
Daniel Veillard      | Red Hat Network https://rhn.redhat.com/
veillard@redhat.com  | libxml GNOME XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/