TDS / Unix CD work

[Ulrik Vieth]

Sebastian wrote:

> After talking vaguely about this for some time on and off, and being
> aware of other similar projects, i have decided to try and go for a
> new Unix teTeX CD within the next 3 months regardless. This would be
> jointly sold/approved by TUG, UKTUG and GUTenberg, if they formally
> agree (I haven't formally asked yet :-}), or by me personally if they
> dont...

First question: what about DANTE (and NTG and whatever other group)?
No, I'm not a DANTE official, just an ordinary member, but I wonder
why you just mention those three groups above.

> Naturally, i want as much help and advice as i can get on this, so who
> better than the veterans of the TDS? Obviously anyone who has worked
> on the 4AllTeX CD will have very valuable opinions about what should
> be on the wishlist.

Next question: I recently discovered that there exists or existed 
a TEX-CD mailing list at listserv.shsu.edu, whose mail archives seem
to indicate that it was active only briefly from December 93 till
March 94, but has been idle ever since the NTG went ahead to produce
their 4allTeXCD (which I personally find a very messy) while most 
of the Unix people started discussing the TDS and took 1 1/2 years 
to finish that.  

Wouldn't it be a good idea to reactivate that mailing list for your
project?  AFAIK, it's open for subscribtion by anyone (I've tried it
recently) while the TDS mailing list is closed and both list-owners
are unavailable.

>    a) to provide a ready to run mountable Unix file system 
>       for a variety of platforms
>    b) to provide an ISO-9660 compliant TDS runtime 
>       tree as large as we can make it, with text files in Unix format
>       (or LF line separators, not CR/LF)

OK, sound's fine so far.  But might I also suggest using RockRidge
Extensions to allow for those files that don't (yet) fit in the 
8+3 name space?  I don't know about the technical impliciations, 
but most of the Linux CDs have it, and it works just fine without 
even thinking about it.  Besides, file names like dvips.info-{1,2,3} 
or kpathsea_toc.html wouldn't travel too well under 8+3 truncation 
and we're aiming especially at Unix systems.

> What Not to Do
> --------------
[...]
>  - Suggest waiting for web2c 7.0
That's what I was about to suggest concerning MetaPost (see below)

> What to Do
> ----------
[...]
> b) get and install a virgin teTeX distribution if you don't have it
... and get more disk space if it's all occupied by texk-pretest...

> Conclusion
> ----------
> Any comments? Suggestions? Feel free to circulate this to anyone you
> think would like to join in.

Some suggestions:

1.) Try to preserve the original file dates/modification times as
    much as possible, i.e. use cp -p or tar when arranging files 
    on the local filesystem.  Likewise, when ftp'ing always get 
    the whole directory as tar.gz not individual files, since tar
    will preserve the time stamps while ftp doesn't.
   
2.) Add lots of documentation in texmf/doc, at least more than
    there is presently in the teTeX distribution.
    Some suggestions: the latest UK TUG TeX FAQ, comp.fonts FAQ, 
    the TDS draft (so that people can understand why things are
    organized as they are), tutorials such as lshort2e/lkurz2e, 
    other useful stuff in CTAN:help or CTAN:info ...
    
3.) Add index.html files to the texmf/doc tree, so that uses can
    point their favorite WWW browser to the mounted CD and find
    the documentation by a clicking and having the appropriate
    viewer (xdvi, ghostview, whatever) invoked automatically, 
    no matter what file format.  I suspect that many people tend 
    to ignore the documentation if it's difficult to find, but
    what could be easier than WWW.

    Caveat: Collecting the raw data for index.html files is easy 
    using the MakeHTML script in teTeX.  But making these index 
    files pretty by adding the titles and a little description 
    can be a lot of work, though.


> Package: A much larger set of PostScript metrics
> Who: SPQR

I'm skeptical about that.  Anything that goes beyond the basic
35 and a few free widely-used fonts probably won't do any good
for a large pecentage of users that don't have extra font CDs.
Besides, while running from the mounted CD is one goal, I think
it should also be easy to copy packages to disk selectively 
if one doesn't want to run from CD.  In that case, a huge font 
tree might not be what everyone wants, or at least it would
make it more difficult to copy some subtrees of the texmf tree.

> Package: aucTeX and lacheck
> Note: location?

In case of doubt, I'd suggest an emacs/lisp tree at the top level.
Still, users running from the CD would have to set their load-path
appropriately or copy an edited version of tex-site.el to their 
site-lisp directory on disk.  The location of the info files 
shouldn't be a problem, though.  Just put them in the teTeX info 
directory.

BTW, are there other non-standard Emacs packages that would be
of general interest on a TeX CD?  Anything like html-helper-mode 
or other HTML/SGML stuff?

> Package: latex2html
> Note: a *real* can of worms. i dont even want to think about it

While latex2html might be difficult, I'd suggest to include
texi2html, especially since Texinfo is already there in teTeX.  
It does require Perl (which we might assume to be available on
many Unix platforms), but as far as I know it works just fine 
with both Perl4 and Perl5.

We might also include texi2html-converted versions of Texinfo
manuals, i.e. Kpathsea, Dvips, Eplain, AucTeX, TDS, ...

> Package: metapost
> Note: needs a source patch, executables, and a diff for texmf.cnf

Well, I guess I would be the ideal candidate for that, should 
I allow myself to be drawn into it.  However, as I mentioned 
recently on the teTeX mailing list, I'd really hate to go back 
to the old version and touch up the MP-0.63/teTeX-0.3 version 
for MP-0.631/teTeX-0.3.3 (with additional bug fixes for MP).
That's why I was about to suggest waiting for web2c-7.0 which
will have it by default (including several minor impovements).
Anyone else who wants to do it?

> Package: omega
> Note: i have asked John and Yannis

Is Phil going to suggest e-TeX 1.0 as well?  Bernd already had 
a change file for web2c-7.0, I believe, and I think it would 
be much easier to include than Omega, since it doesn't change
the architecture.  (I had tried e-TeX once with teTeX-0.3 
when it was first announced, but I never pursued that further.)


Other packages of potential interest:

- filehdr (and checksum)
  consist of an elisp package and a small C program

- psutils: psbook, psselect, psnup, ...
  most of them C programs, some might require Perl

- cweb (or cwebx), wmerge, tie, other LitProg tools?


OK, so far these are just some comments and ideas. I'm a bit
reluctant to really offer my help on anything specific since
I've already spend too much time on various projects recently 
and should better concentrate on getting some real work done.

Anyway, it looks good so far and I wish the project success.  
I hope it'll really get done this time and I'm looking forward
to the first usable TeX CD that neither suffers from a messy 
organization nor from having everything packaged up making
it difficult to get at anything (such as having the teTeX
binaries for all platforms packaged in a huge 20 MB zip file 
only to preserve the long file names of the tar.gz files).

In case of doubt, I'd suggest that less is more, i.e. better
concentrate on the well-known standard stuff and have that
well organized rather than trying to fill the CD to the max.

Cheers, Ulrik.


P.S. Last question: Any idea about the price of the CD?  
The latest CTAN CD produced by DANTE in cooperation with 
Addison-Wesley sells  at DM 50,- with book or DM 10,- 
(CD only) for DANTE members.