Message-ID:  <m0xL76z-000OWXC@dcarlisle.demon.co.uk>
Reply-To: Mailing list for the LaTeX3 project
              <LATEX-L@RELAY.URZ.UNI-HEIDELBERG.DE>
In-Reply-To:  <1622-Tue14Oct1997113530+0100-s.rahtz@elsevier.co.uk> (message
              from Sebastian Rahtz on Tue, 14 Oct 1997 11:35:30 +0100)
Date:         Tue, 14 Oct 1997 14:34:13 +0100
From: David Carlisle <david@DCARLISLE.DEMON.CO.UK>
Sender: Mailing list for the LaTeX3 project
              <LATEX-L@RELAY.URZ.UNI-HEIDELBERG.DE>
To: Multiple recipients of list LATEX-L
              <LATEX-L@RELAY.URZ.UNI-HEIDELBERG.DE>
Subject:      Re: frontmatter 98
Status: R

Sebastian

> If any part of fm98 ... uses keyword=value, then we might as use it
> for *everything*.

Well I agree with that (and probably agree with using that kind of
syntax in general) but I think that in the *everything* I quote above
you meant `everything in the front matter'.

One reason I didn't use KV in the fmatter package I sketched before
was the belief that if you can pursuade authors that they should
use
\author{MothersMaidenName=Willingham,forename=David,
 MiddleNameButIDontUseItMuch=Paul,surname=Carlisle}

then they might want to start using

\section{level=C,tocheading=this,runninghead=that,maintitle=the other}

or 1001 other possible extensions. In that case the project is no
longer about coming up with a frontmatter specification but re-writing
the whole of latex to use named arguments. That isn't a bad idea,
but it's probably a different project, one in fact that is not totally
unrelated to this list...

In a similar vein

> the keyval approach has the great
> advantage that she need not commit herself immediately to what all
> those tags are

You don't have to use KV to get that. Using the undef.sty package I
posted, the fmatter classes can define any new command. Any class that
does not define that command will skip past the undefined command (and
any {} or [] arguments) with just a warning message, no error
generated).

However to assume (for now) we go with a keyvalue syntax, you say

> We don't want to repeat address Y, do we?

If this was only aimed at TeX there would be an obvious solution since
TeX has quite a powerful macro facility built in:

\newcommand\addrone{a\\long\\full\postal\\address}

\author{name=Me, address=\addrone}
\author{name=You, address=\addrone}

If you allow TeX expansion on the key values, then you can have any
amount of saving of keystrokes without the frontmatter system building
in an explicit cross referencing scheme.

Possibly though you would want to discourage the use of such general
macro definitions in the front matter and restrict the possibilities,
either so the frontmatter could be viewed as a general database markup
without running it through TeX, or because you just want to easily
check the author isn't doing anything too horrible
(\renewcommand\section......). In that case some (all?) keys would
need some cross referencing scheme as you sketched.

> do we simply add more and
> more keys to \author, to cover such situations?

I don't think you can catch all uses. You will need some kind of
generic `note' key analogous to the dreaded \thanks. If every class
comes up with a new specific key for special combinations of temporary
address then it will once again become hard to share documents between
journal classes. For most purposes it would be sufficient to have
a `temporary address' key with some kind of attatched note to say
`study leave' or `on a visiting KJFJD fellowship' or whatever.

If one class defines a new key for `addresses of KJFD visitors'
then other class files are not even going to recognise that as an
address at all, and will have to just ignore it. Whereas a general
address key with an attatched note is more useable. This is assuming
you are trying to make documents out of this. If you want to do a
database lookup to find all KJFD scholars over the past decade then
having a key for it would help, rather.


David