On Mon, Mar 06, 2006 at 01:46:48AM +0100, Danilo Šegan wrote: > Yesterday at 11:28, Daniel Veillard wrote: > >> It might cause problems for xml2po as well since translations might > >> want to use a different "my-entities.ent" (of course, I'd need to fix > >> xml2po to support "--keep-entities" as well, which is probably why I > >> didn't notice this sooner). > > > > Relying on parameter entities substitutions for localization doesn't > > sounds a good idea to me, very fragile and relies on the most complex > > part of the XML-1.0 specification. For example a parser operating in > > non-validating mode of XML-1.0 is allowed to drop any declaration occuring > > in the internal subset after a parameter entity reference, and will do > > so without warning, fragile, very fragile ... > > I agree, that's why gnome-doc-utils even uses "-e" (expand all > entities) parameter for xml2po in default document generation rules > (this is also necessary so localisation would be possible in all > cases). But this is what users seem to be doing anyway :(
Well preserving general (i.e. non-parameter) entities should be reliable in libxml2 wherever they happen, but it's an editing toolkit property, for example other parsers based on SAX will for example loose entities when they occur in attribute content. Daniel -- Daniel Veillard | Red Hat http://redhat.com/ [EMAIL PROTECTED] | libxml GNOME XML XSLT toolkit http://xmlsoft.org/ http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/ _______________________________________________ xml mailing list, project page http://xmlsoft.org/ [email protected] http://mail.gnome.org/mailman/listinfo/xml
