Jeff Turner wrote:
>>Mixed text works best when it is used for, well, mixed text.
...
> Btw, the doc-v12 DTD (and XHTML) both have a
> 'flow' data type which mixes block-level and inline elements ('flow' is
> used in li, dd, etc).

That's not really critical, usually. It's still all content.
You were mixing text content and a sort of out-of band data.
This *is* critical.

Note that a DTD would have a content model of
  <!ELEMENT question (#PCDATA|elaboration)*>
which would also validate
  <question>
    Bla bla bla
    <elaboration>stuff</elaboration>
    more bla bla bla
    <elaboration>more stuff</elaboration>
    even more bla bla bla
  </question>
Ouch!

> Extra metadata for FAQs would be great.

Oh, certainly. The problem with most metadata is that it tends to
fall into disuse unless automatically managed.

>  - Date the FAQ was added, rendered as a "New!" icon for FAQs that are
>    less than X weeks old.
>  - Version of software the FAQ applies to, allowing automatic
>    filtering-out of irrelevant FAQs.
>  - Author, mailing list ref, etc.

Neat ideas. But don't expect this to be added consistently
unless either
- committing fails unless plausible values are entered (or
   send nag mails, or something else obnoxious), or
- there is a dead-easy to use app for updating FAQs which takes
   care of the metadata, and only a minority uses more-or-less-
   plain-text editors and gets shouted at if the valuable
   metadata is missing...
Note that if there is even only a temorary problem with keeping
the metadata accurate, it no longer provides added value, and
the whole idea usually collapses very quickly.

You wouldn't belive how often I experienced this already.

> We can do this in faq-v20 with a <meta name="..."> ... </meta> tag
> allowed inside <faq>. 

*Cough*! Do you want people asking why the encoding they
added to a FAQ "doesn't work"?

>>BTW will you rename the FAQ <part> to <faqsection>?
> I have done so for faq-v20a.

Finally!

J.Pietschmann