Message-Id: <9212191856.AA03499@sc.zib-berlin.dbp.de>
Reply-To: LaTeX-L@vm.urz.Uni-Heidelberg.de
Date:         Sat, 19 Dec 92 19:56:42 CET
From: Frank Mittelbach <MITTELBACH@MZDMZA.ZDV.UNI-MAINZ.DE>
Sender: Mailing list for the LaTeX3 project <LATEX-L@vm.urz.Uni-Heidelberg.de>
To: Multiple Recipients of <LATEX-L@vm.urz.Uni-Heidelberg.de>
Subject:      overloading of function names
Status: R


Now that my email situation is becoming slightly better I'm faced with
a different problem: a huge amount of mails to answer (or not to
answer :-). I guess I pick them up at random, they are just too many.


> Paul Taylor's opinion, whatever its validity (which I do not question),
> has nothing to do with the original question, which is what happens
> to ' (and `, I suppose) inside the tabbing environment.  For whatever
> reasons, having nothing to do with EBCDIC or 7 bits or anything I consider
> it bad form to change a definition of that sort inside an environment.
> It creates mysterious errors, makes macros fails and is just poor programming.
> It is all very well for a person who never uses the tabbing environment
> (till recently he never used latex) to say not to bother, but for those
> who do, it can be a real mess.  It was an error to do this and latex 3 is
> the time to repair it.

A lot of people commented on the original question by Dominik and the
discussion was floating even into the uni-code corner (remember, we are
not trying to rewrite TeX in this project, do we?)

I think the question boils down to: is overloading of operators
something sensible or should it be avoided completely?

To give some other example where this is a problem in the current
latex: \\ is used for a line separator in various places, for example
in raggedright, tabular, and, and, and. Now everybody who had done
some latex support will have been faced with the baffled user who
tried to use \\ inside a p-column of a tabular to denote a linebreak
rather than a column sep.

In my opinion Michael Barr (I think he wrote that piece above, right?)
is right by saying that for ltx3 it is time to repair that mess.

The question is how?

One extreme is to avoid overloading completely but I don't think that
is really the right decision. I understand overloading as something
which is done to a) ease keying in a document, but also sometimes to
b) help clarifying the issues. For example, it is nice to think of \\
as a line separator since that superficial view is correct until you
try to combine different uses.

Thus I would propose to remove overloading on the basic level (ie
different functions get different names), eg \newline, \rowsep,
\acuteaccent, \graveaccent, etc. together with a standard mechanism
for mapping such basic functions to other names in certain contexts.
This would mean that some shorthand functions could change meanings if
the style designer or the user (in the preamble) decides this, but
there would be always the possibility to resolve conflicts in
situations where such overloading would be ambiguous and there would
be the possibility to adjust the conventions to personal taste in a
standardizied way without making the documents unexchangeable.

this could, for example, be done in ``input convention style files''
similar to the input conventions styles for the current ltx that
define shorthands for ease of keying (like german.sty defining various
double quote short hands, eg "a "s etc.).

In contrast to the current situation we will have in ltx3 a standard
way of mapping such functions to that the combined use of such files
would be possible (which is often not the case for such extension like
german in ltx209, because they change catcodes which other styles
don't anticipate).
By the way, a similar mechanism for short-hand strings is already
developed.

           -- my $0.02 worth of this old discussion

Frank

ps all of you a merry christmas and a happy new year