On Jan 3, 2012 7:29 PM, "John Cowan" <cowan@mercury.ccil.org> wrote:
>
> Henri Sivonen scripsit:
>
> > > 2) Clearer story on how to use the ISO/MathML entity sets along with XML
> > > Schemas, or with no schemas. I often hear "XML Schemas does not support
> > > entities, so I cannot use entities with XML Schemas" or vice versa.
> > > Preferably, build the default mappings of the entities into XML.
> >
> > Adding entities to XML without DTDs would break the compatibility of
> > new documents with old parsers. I think entities alone are not worth
> > such a discontinuity.
>
> Not if it's done at the character encoding level. The proposed utf-8+names
> encoding provides for that: see http://www.tbray.org/tag/utf-8+names.html .
...
> Totally XML-conformant, and any XML processor can implement it if it
> wants to. Then it's just a matter of putting a correct XML declaration
> on your documents. You don't even need your editor to implement the
> encoding: a regular UTF-8 editor is fine.
The key this isn't conformance but breakage in existing software. To the extent existing software properly treats unknown encoding names as fatal errors, it doesn't matter if UTF-8+names manages to make itself spec-wise conforming.