Message-Id: <9212081408.AA06684@sc.zib-berlin.dbp.de>
Reply-To: Mailing list for the LaTeX3 project <LATEX-L@vm.urz.Uni-Heidelberg.de>
Date:         Tue, 8 Dec 92 15:06:27 +0100
From: David_Rhead@VME.CCC.NOTTINGHAM.AC.UK
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:      Papyrus: copy of mail I've sent to Dave Goldman
Status: R


Here's a copy of the mail I've sent to Dave Goldman (the Papyrus chap).
---------------------------------------------------------------------------
Thanks for your mail, which I have forwarded to LATEX-L.

As regards listservers, the Heidelberg one is OK. Thus, send a message
either to
   LISTSERV@DHDURZ1.EARN
or to
   LISTSERV@vm.urz.uni-heidelberg.de
saying
   SUB LATEX-L Dave Goldman
[LATEX-L is a general discussion list for LaTeX 3.  There may be a case for
hiving discussion of bibliographic issues off onto some other list.  But in
the meantime, LATEX-L will give you the flavour of the project.]

Traditionally, LaTeX has inter-worked with the public-domain BibTeX, and I
suppose that would set a precedent for how LaTeX might inter-work with
other things.  Ideally, therefore, you really need a reply from someone who
* is familiar with the sordid details of how LaTeX and BibTeX inter-work
  now
* is familiar with how LateX/BibTeX may inter-work in the future
* can see how the details might be adapted/generalized so that LaTeX might
  inter-work with bibliographic software other than BibTeX.
I don't think that I'm the "someone", but I'll answer as best I can so that
at least some of your points get dealt with promptly.  (I'll forward a copy
of this to LATEX-L so that the project-managers get an opportunity to
correct any wrong impressions that I may have.)

>    It sounds like the major area of concern that you've identified so far
>    is that of included subfiles.

Yes, I mean situations like when a book is made up of chapters.  E.g.,
someone doing a thesis might have a root file root.tex
  \documentstyle...
  \begin{document}
     \include{intro}
     \include{my-work}
     \include{concs}
     \include{back}
  \end{document}
and supply 3 chapter files intro.tex, my-work.tex, concs.tex and a file of
back-matter back.tex.  The command the end-user issues is [probably]
   latex root
In current LaTeX, there are auxiliary files root.aux, intro.aux,
my-work.aux, concs.aux, back.aux, and between them they take care of
bibliographic citations, typically from the chapters to the back-matter.
If BibTeX is being used, the end-user goes
   bibtex root
   latex root
   latex root
to get references taken from the user's database file and put in the back
matter, and citations inserted in the text.  In essence it may be similar
to "the Papyrus workaround" you mentioned, but the "running through the
chapters" is done automatically, since root.aux "knows about" the chapters.
My understanding is that this process is under review for LaTeX 3 (e.g.,
there might be just one .aux file).  E.g., if one way of doing it makes it
easier for co-working with proprietary bibliographic software than another,
my impression is that, other things being equal, the "easier with bib
software" option could be chosen.  Perhaps it could be done so that it
doesn't look like "a workaround" to the end-user.

>     Tracing down included file trees would not be difficult for us to do,
>     in principle, if it would actually be of significant utility.  In
>     practice it could get a bit snarly, though, so I'd rather be convinced
>     of its usefulness before offering you any promises!

In LaTeX jargon, perhaps it's \input files that would be snarly.  As I said
above, the root file "knows about" \include files.

>     My understanding of LaTeX is that it is a "mark-up" specification
>     independent of the actual text editing software in use.

Yes.  For LaTeX 2.09, see "LaTeX: A Document Preparation System" by Leslie
Lamport, Addison-Wesley, 1986.

>     Unless this has changed, and there is now a *program* named LaTeX or
>     something, then there are limits to how you could design an
>     "interface" that would work to connect all text editors to all
>     bibliographic programs.  When you say "Perhaps a command xxxx could be
>     defined such that...", I don't understand who is the recipient of this
>     command: the text editor?  the bibliographic program?  the operating
>     system?

>     If you are simply talking about a shell command, then you might as
>     well just launch the bibliographic program and point it at the LaTeX
>     file.  You'll have to provide additional parameters anyway
>     (bibliographic style, mainly), and for now these are very specific for
>     each bibliographic program.

I suffer from not yet having had "hands on" experience of proprietary
bibliographic software, and so guessing at what may be possible.  (I hope
to get some in the next few months, but I don't have it yet.) I'm guessing
at:
* a shell command that the user would issue instead of the bibtex command
* rather than "pointing it at" the .tex file, there may be advantages in
  pointing it the .aux file (since the .aux file is shorter, and will
  contain things like \citation{smith-92} that say what references are
  required)
* from the end-user's point of view, then, instead of going
     bibtex root
  I wonder if they could go
     xxxx root
  for some xxxx, where (in the Papyrus case), xxxx invokes Papyrus in such
  a way that it "points at" root.aux and the user ends up with a bibliography,
  in the style they desire, in root.bbl (as they would have done from BibTeX).
To do better than guesswork, I think it needs someone who does have "hands
on" experience of Papyrus talking to someone who knows the range of
possibilities of how LaTeX 3 might be organised to interact with bib
software.

>    PAPYRUS can be instructed to spit out a pretty good imitation of a
>    BibTeX file.  So right now you can use PAPYRUS to suck in references
>    from a CD-ROM and turn them into BibTeX.  (PAPYRUS can also import
>    BibTeX files.)

Such considerations may have implications for BibTeX 1.0 (or, to be more
precise, for the associated .bst files).  E.g., if people such as yourself
had a common analysis of structure (analogous to the fields/entries in
Appendix B of the LaTeX manual), one could imagine .bst files that
implemented the same analysis [probably a little different from that in
Appendix B], thus making conversion better than "pretty good").  ["Imagine"
may be the operative word.  I'm not sure how agreement could ever be
reached about what this structure should be!]

>     At the moment, all of us proprietary guys are running on DOS and/or
>     Macintosh platforms.  (A couple of us also do VAX-VMS.) I'm not aware
>     of any notable proprietary bibliographic software for UNIX (though
>     some of my colleagues may be working on that).  My impression is that
>     TeX systems, though available for DOS and Macs, mainly exist on UNIX
>     platforms.

(La)TeX runs on a wide variety of computers.  There are good public-domain
implementations for DOS and Macs.  My impression is that (La)TeX users are
following the "downsizing" trend.  Thus, although there is certainly
(La)TeX on Unix boxes, (La)TeX on DOS and Mac has a sizeable (and increasing)
following.

Indeed, if you want to get a TeX with which to experiment on co-working
with Papyrus, you'd probably find it convenient to get
either emTeX for DOS
or     ozTeX for Macintosh.
(Both public domain.)

>     So although I am eager to get on with this discussion, how relevant
>     will it be?  Are UNIX folks going to be willing to pay money for
>     software?  Are PC and Mac owners going to switch to LaTeX-based word
>     processors?

I guess that the LaTeX niche will be:
*  scientific work
*  top quality typesetting, e.g., for books, journals,
   conference-proceedings.
I don't imagine that Wordperfect and MS-Word will go away.  But for "proper
publishing", people may use a PC or Mac wordprocessor/editor to create
files for input to (La)TeX.  So I'd guess at a continuing niche, and that
(since LaTeX 3 will be more tuned to what publishers want than LaTeX 2.09),
the niche may be more substantial than the current niche.

I can't really speak for the world's "Unix folks"!  I'd guess that some
would stick to public-domain stuff, e.g.  BibTeX.  Others might want common
software on all their organisation's platforms, hence:
* choose a proprietary system for their PCs (so they get the interface to
  databases, and get something that works with wordprocessors for things
  outside the LaTeX niche)
* then go for the same thing on their Unix boxes so that they can interchange
  databases.


                                                                David Rhead