Message-Id: <9311032006.AA25036@sc.ZIB-Berlin.DE>
Reply-To: Mailing list for the LaTeX3 project <LATEX-L@vm.urz.Uni-Heidelberg.de>
Date:         Wed, 3 Nov 93 21:04:00 +0100
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:      nf prefix and statement on LaTeX2e compatibility
Status: R


I think it might help to give a bit of history as well as some
statement of intentions concerning the LaTeX2e release.


NFSS2 FILE NAMES:
=================

First the more easier part ...

 When NFSS2 was designed many people involved thought it would be a
good idea to the have a set of uniform naming convention since it
would (at least for a long time) be ``just another'' FSS sitting side
by side at least with the plain TeX times.sty or some lonts-based
times.sty. so nf as a prefix came into life.

this situation has changed very much with the decision to uniform all
such latex flavors and release LaTeX2e that is able to run all
variants under a single format (the has a few consequences on NFSS2 as
well, see below). thus if the public opinion is strongly for stripping
"nf" i do have no problem with that, although AW will not like it that
i change the LaTeX Companion once more a few weeks before printing.


> From: Francisco Figueirido <figuei@LUTECE.RUTGERS.EDU>
>
> Anyway, I find it more disturbing that the user-interface to NFSS2
> (especially regarding mathematical typesetting) has changed so much with
> respect to NFSS1. Luckily we haven't yet used NFSS too much ...

I'm not sure that i can follow here, are you refering to the declaration
commands like \DeclareMathAlphabet?

I agree that this this part of the interface has changed drastically,
although there was not much choice if you want to provide `encoding'
as an important additional attribute (although perhaps not so much for
math at the moment). Anyway, while nfss1 was (as far as math was
concerned) a horrible hack nfss2 actually gives you a clean and usable
interface for additional symbol fonts, so that it now isn't a problem
any longer to integrate your personal symbol font and/or to typeset in
completely different math fonts. Last not least, nfss2 will accept the
nfss math commands and convert them to nfss2 (with a warning).

as i said above, due to the decision to integrate nfss2 into latex it
will undergo a number of small changes outlined below.


COMPATIBILITY
============

Joachim, I think, said:

> We have hundreds of long-living documents that use constructs like
>
>     $({\it colon}, {\it space})$
>
> all over the place. If they won't work with a new system, I won't
> switch. And I will recommend the same action to _everybody_ in a
> similar situation.

The above wouldn't work under nfss1 (without using oldlfont) and also
not under nfss2 (without using nfoldfnt) the way you find it currently
on the servers. However,

    it will work under latex2e (and not only under compatibility mode)
    ------------------------------------------------------------------

you see, the is already the first change that will happen to nfss2 in a
few days, so why not change the file names just before reaching home,
if everybody thinks that is better.

> I don't have a problem, if the constructs above won't work with
> documents starting with \documentclass. But if they start with
> \documentstyle, they *must* be compatible. The
> installation-effort--to--use-enhancement ratio is too low otherwise.

so let me state here what kind of backward compatibilities we have:


Document level:
---------------

- Sources starting with \documentstyle will work as under LaTeX209.

  Possible exception: if they use undocumented commands coming out
  lfonts.tex in the source, there probably will be undefined commands
  if that really turns out to be a larger problem one can make the
  compatibility mode even stronger by trying to emulate `one!' of the
  many lfonts existing in the world. this is techincally not a problem
  but eats a lot of space. Actually at this point we are in the area
  where individual sites had different commands available.


- Sources for LaTeX2e will start with \documentclass so that they
  don't have to load all kind of compatibility internals but can
  instead load additional features.


Styles:
------

 Styles in LaTeX2e are divided into two classes:

       document classes    and      packages


Classes are those that appear in the mandatory argument of
\documentclass (\documentstyle). Packages are those that appear in a
\usepackage command (or the optional argument of \documentstyle).

- document style files (like darticle.sty) will work as before under
  LaTeX2e if their extension is renamed to .cls. In addition one might
  have to add definitions (trivial) for \rm etc. see below (they can
  be copied from the standard classes coming with latex2e.

- other style files, eg epic.sty will work unchanged with latex2e if
  used in a \usepackage command, (or in the optional arg of
  \documentstyle for and old document).

  Execption: as for the source documents: if the styles refer to
  lfonts.tex commands they will probably need a bit of updating.


However, latex2e give a much more flexible option concept for
package/class (style) writer which of course can be used only by
new or updated styles.


To sumarize:
============

The LaTeX2e is meant to embrace all such TeX flavors so that is not
necessary any longer to have four different latex versions on a site
to be able to process documents (without changes) coming from other
sites.


Changes to NFSS for integration:
================================

To the problem of different latex's having different ideas about \rm
\it in math (yes/no) and so on, the following is going to happen:


    NFSS2 (now)		LaTeX2e

    \rm			\rmfamily
    \bf			\bfseries
    \it			\itshape
    .......		.....    (you get the idea)

    \reset@font		\normalfont   (\reset@font will still be usable internally)

    \textem{...}	\emph{..}
    \textit, and the others  stay but got full italic correction facility


That completely frees \rm etc from any meaning. In other words,
LaTeX2e format (!) does not give them a meaning at all. Instead all
those command will be defined by the class files.

All standard classes will define such cmds to work as in 209 although
the updated documentation will suggest not to use them (or even
encourage to user preference definitions in the preamble)

There will be a few additions, for example

DeclareMathSizes{<text-size>}{math-text-size}{math-script-size}
                             {math-scriptscript-size}


> Also, if your documents started with a (self-documenting) command such as
>
>   \TeXformat{LaTeX}{2.09}{<date>}
>
> it would be easier to preserve compatibility.  But the \TeXformat
> command would have to be recognized by the format file, of course.
>
> Michael Downes                              mjd@math.ams.org (Internet)
>
>
> > Also, if your documents started with a (self-documenting) command such as
> >
> >   \TeXformat{LaTeX}{2.09}{<date>}
> >
> > it would be easier to preserve compatibility.  But the \TeXformat
> > command would have to be recognized by the format file, of course.
>
> Yes, that's ok -- let's introduce this with LaTeX2e. As I wrote, I can
> live with the fact that the `plain way' does not work anymore in a
> LaTeX2e or LaTeX3 document. But older LaTeX 2.09 documents should
> still be valid input and should produce results which are not too far
> away from the previous ones.


It is introduced, with a slightly different name if you need it:

\NeedsFormat{LaTeX2e}[1993/12/24]

I hope that we can get this into amstex as well and perhaps even into
plain (in some primitive definition) so that a file that really needs
a certain format can ask for it and the user gets a sensible error
message, although


! Undefined control sequence.

  ... \NeedsFormat
		  {LaTeX2e}[1993/12/24]

l. 3

isn't too bad.


I hope this has clarified a few things.
This is not meant to be something like a WordPerfect upgrade.

Let me finish with some excerpt from a mail by Leslie Lamport:

   REASONS FOR LaTeX2e
   --------------------
   There are two primary reasons for introducing a new version of LaTeX:

   1. To incorporate NFSS into standard LaTeX.

   2. To add to standard LaTeX graphics functionality that is currently
      obtained with driver-dependent macros.

   As long as we are introducing a new version, we will take the opportunity
   to add some simple features that should have been in version 2.09.

   Although LaTeX-3 will eventually do all this, we feel that it is
   important to achieve these goals soon to prevent the proliferation of
   mutually incompatible dialects of LaTeX 2.09.


   GUIDING PRINCIPLES
   ------------------
   Since LaTeX2e is meant to tide us over until LaTeX 3 is released, it
   is important that it be viewed as a minor modification to LaTeX 2.09.
   Thus, the following two guiding principles are to be followed:

    I. Unmodified version 2.09 input files will produce the same output
       with LaTeX2e as with version 2.09, except when it is absolutely
       impossible to reconcile this with NFSS.

   II. All new features of LaTeX2e will conform to the conventions of
       version 2.09, making it as easy as possible for current users to
       learn to use them.


Frank Mittelbach
LaTeX3 Project