Pav Lucistnik <[EMAIL PROTECTED]> wrote in <[EMAIL PROTECTED]>:
pa> Right, haven't noticed this. Do we have a language-dependent place for pa> this? Does xsltproc have commandline argument to specify encoding pa> perhaps? Or do we have to copy the .xsl around? Please put en/share/sgml/templates.usergroups.xsl. See: http://www.freebsd.org/cgi/cvsweb.cgi/www/ja/share/sgml/templates.usergroups.xsl?rev=1.1&content-type=text/x-cvsweb-markup pa> > And your commits seem inconsistent; changes for www/89380 replace pa> > &#xxx; with &foo;, but entities.dtd adds such entities in &#xxx; form. pa> > Which one do you think better? pa> pa> Whatever is easier to read and write by humans. &foo;s clearly wins. This is not only a style issue. What we must consider is that the &#xxx; form can depend on a specific encoding (this is always UTF-8 in the XML spec, though) and some web browsers do not handle them correctly. The entity references should be &foo; form in the *resulting* HTML files wherever possible. The XHTML entities are actually defined in the &#xxx; form (you can see them in $PREFIX/share/xml/dtd/xhtml/xhtml-lat1.ent, for example), so we never put them in the resulting HTML files in &foo; form even if we add entities.dtd and use &foo; in an XML file. I think a concept of "sdata-as-pis" used in osx(1) is a good idea for XML files in our www and doc tree. Redefining all of ISO 8859 entities in the SDATA PI form and adding a template for that, we can put them in the resulting HTML files in the &foo; form. -- | Hiroki SATO
pgpKWr41M9fgi.pgp
Description: PGP signature
