User-Agent: Thunderbird 2.0.0.22 (Windows/20090605)
MIME-Version: 1.0
References: <4A7921CF.5020803@morningstar2.co.uk>           
            <4A8EC449.4040509@morningstar2.co.uk>           
            <19088.5371.517713.176151@morse.mittelbach-online.de>
            <F935973B-32E8-4FD2-9B02-6F55FCB73674@gmail.com>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Message-ID:  <4A92765A.3040807@morningstar2.co.uk>
Date:         Mon, 24 Aug 2009 12:15:38 +0100
Reply-To: Mailing list for the LaTeX3 project
              <LATEX-L@LISTSERV.UNI-HEIDELBERG.DE>
Sender: Mailing list for the LaTeX3 project <LATEX-L@LISTSERV.UNI-HEIDELBERG.DE>
From: Joseph Wright <joseph.wright@MORNINGSTAR2.CO.UK>
Subject: Re: xparse
To: LATEX-L@LISTSERV.UNI-HEIDELBERG.DE
In-Reply-To:  <F935973B-32E8-4FD2-9B02-6F55FCB73674@gmail.com>
Precedence: list
Status: R

Will Robertson wrote:
> Point B is something I think I'd like to work with more in the future,
> and as layers 0 and -1 start separating (see parallel thread) it may
> become more useful. Having said that, at this stage it's probably of
> little use so I don't expect much to happen with
> \DeclareDocumentCommandInterface at this point in time.
> 
> (I sort of feel that if it's to be used, it should be pervasive -- the
> benefits of separating interface/implementation are diluted if only a
> subset of commands are defined in such a way.)

I'd wondered about implementing \DeclareDocumentCommand so that it
always did \DeclareDocumentCommandInterface and
\DeclareDocumentCommandImplementation. However, I'm not sure this really
works that well. As I've already pointed out, we still rely on the two
parts having the same number of arguments.  That will fall down if we
imagine something like:

\section*[short]{long}

versus

<sectiontitle numbered="false",shorttitle="short">
  long
</sectiontitle>

or some such XML-like input, at least without an intermediate layer.
-- 
Joseph Wright