X-VM-v5-Data: ([nil nil nil nil nil nil nil t nil]
	[nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil])
Date:         Fri, 6 Jul 90 08:42:32 CET
Reply-To: LaTeX-L Mailing list <LATEX-L%DHDURZ1@unido.bitnet>
From: PZF5HZ@RUIPC1E.bitnet
To: Rainer Schoepf <BK4@DHDURZ1.uucp>
Status: R
X-Status: 
X-Keywords:                  
X-UID: 165

Don raised some important topics about makeindex and I like
to add a few remarks:

    (1) despite what Nelson says, makeindex should be
    in WEB.  Somewhere, while I was at Texas A&M I read
    a comment saying "makindex is a fine program, but
    only runs on Unix since it was written in C." ....

I agree with Don that we should try to provide all TeXware in Web
format. Not because it is not possible to produce good coded
programs but because I think it is very important to provide
an environment which is likly to run on all target machines.
Using Web2C translators it is fairly easy to use Webware
in the Unix world and Web itself ensures that the program are
likely to run on all other machines. Thinking of commercial
(industrial) environments we often have to deal with IBM mainframes
where you might find a Fortan, Cobol, PL1 and Pascal compiler
but often no C compiler (expensive).

    (2) Speaking of documentation, well, um. Makeindex, as Nelson
    distributes it, has three documents:
          ...
       makeindex - a general purpose, formatter-independent index
          processor (apparently by Pehong Chen). This is the only
          documentation of the ist format. However, the options one
          has for the documentation formatting are: GNU infor format,
          Unix man page, nroff output, VMS help and PostScript output
          from troff. *no* TeX formatting, nor is their a plain text
          file version. No wonder I never figured this stuff out when
          I was at UIC on a CMS machine without PostScript printers.

I have an LaTeX version of this document which was provided by
Reinhard Wonneberger. I suppose that some updates are necessary
but I'm sure that this will be of some help.

    (3) some design flaws: There should be a syntax for indicating
    the ist file to be used in the idx file itself. I'm sure more
    than one person has cursed the day Frank Mittelbach was born when
    they tried typing something like
       latex array
       makeindex array
       latex array
    and that's too bad since it isn't really Frank's fault (well,
    maybe a little ;-) I'm not sure if this can be handled in a
    formatter-independent way, but hey, let's make this feature
    TeX/LaTeX bound if we have to... it is important.

First of all I suppose that there aren't many people arround knowing
this day (at least not under the doc.sty users). But of course Don
is right both makeindex and me could have done better.
Within the last year I had to do a lot of work concerning
automatical index generation and if found more than one important
concept missing in makeindex.

What makeindex calls a style file has nearly nothing to do with
the term style file we know from LaTeX or BibTeX. 90 percent
of an ist file concerns input character translation and not
formating. But input character translation is a property of
the document containing the data. So when the doc.sty uses \index
it use the convention to have the = sign as the actual character
instead of the @ sign because the @ sign is used all the time in the
index data.
So it is only reasonable to request a specific style by writing the
information into the idx file. But I would like to go a step further:

  1) leave the style file for user entries so that the user can
     choose  his / her own special characters.

  2) Allow to reset/change the special characters inside the
     idx file by switching to a specific char map which was declared
     somewhere before inside the idx file. This would allow to
     set up macros for automatic indexing which use their own
     convention for special characters. This is very important
     if you like to combine indexentries from such sources (like the
     doc style) and user entries.

  3) I see no problems to include such a scheme efficent and
     formatter independent into makeindex. The only thing you have
     to do is to add a few more keywords. (The usual meta-meta stuff)

There are many other questionable features in makeindex, for example
sorting which is not properly handled. The same is true for BibTeX
and some efforts should be spend on making the interfaces similar.

Frank