Message-Id: <199402101338.AA20986@mail.cs.tu-berlin.de>
Reply-To: Mailing list for the LaTeX3 project <LATEX-L@DHDURZ1.BITNET>
Date:         Thu, 10 Feb 1994 14:35:48 +0100
From: Frank Mittelbach <MITTELBACH@mzdmza.zdv.uni-mainz.de>
Sender: Mailing list for the LaTeX3 project <LATEX-L@DHDURZ1.BITNET>
To: Multiple recipients of list LATEX-L <LATEX-L@DHDURZ1.BITNET>
Subject:      The difference between option and packages in 2e
Status: R

Mike asked about the difference between options and packages.

> Actually, I expected more feedback on the distinction between "package" and
> "option", but nobody seems to have latched onto the reason why
> "philosophical" appeared in the subject line. Perhaps Paul has hit on a
> possible distinction.  That [size=12pt] is an option, but a4 is a package?
> Please enlighten me someone!

from a pragmatic point of view we have the following simple difference:

  an options is something the class (or package) knows about
  a package is something the class doesn't know about

as a consequence a class can do the right things (in the right order)
if several options are specified but this may not be possible for
packages which may or may not work together (or only i a special
order).

However, this is not the philosophical part of it.
The option mechanism tries to capture the following:

  Documents should be useable with or without the options to a package
  or to a class, ie options conceptually should only modify appearance
  but not adding new strutures.


therefore a4paper is an option and so is leqno but array is not as it
adds new preamble options to tabular and array. [if somebody now
points out that makeidx should therefore be a package and not an
option (as it defines \see) please note that this was an oversight
which is corrected]

the next question is fleqn: is this an option, or a package (since it
defines a new parameter \mathindent). I think it should be considered
an option and the above statement should be extended to

  Documents should be useable with or without the options to a package
  or to a class, ie options conceptually should only modify appearance
  but not adding new strutures. It is acceptable that options define
  additional layout parameters so that the removal or adding of
  options needs adjustments in the preamble of the document (adding or
  removing parameter settings).

That leaves the question of code that only changes the layout but does
not add structure. Eg something that changes article.cls written by
somebody.

Here we do have two possibilities:

- one is to produce a new class file that loads article and modifies
what is necessary, - the other is to write a package nevertheless
although that has the danger that people try to use it with some other
class where it doesn't work.


-------------------------------

however, nothing of this is enforced. It is only suggested. We can't
and wont prevent people to write class/package files with options that
define additional structure but we hope that the general idea will be
picked up and not widely misused.


-------------------------------

Paul said in response to Mike's question

> I don't think it's very natural to call packages .sty.

I do agree and originally we thought of calling them pkg (you can find
traces of this in the docclass code i think :-) However we finally
decided to keep .sty for the simple reason that probably 99% of 209
style options (that define new structure or change layout or both) can
be used unchanged as packages in 2e.  It would have made compatibility
with 209 very very difficult if we had changed that. In case of class
files we decided to switch because of the confusion in 209 that some
files are document styles and some aren't but there wasn't a
distinction.

hope that answers the question a bit

frank