[tex-live] teTeX: no next release

[gnwiii at gmail.com]

On 5/29/06, Peter Flynn <pflynn at ucc.ie> wrote:
> David Kastrup wrote:
> > Zdenek Wagner <wagner at cesnet.cz> writes:
> >
> >> On Mon, 29 May 2006, [iso-8859-1] Frank Küster wrote:
> >>
> >>> Furthermore, I think the best thing we got from Thomas is the
> >>> infrastructure, which has been merged into TeXlive and is still
> >>> maintained by him there.  Work on selecting and updating CTAN
> >>> packages, and in creating useful collections, is probably better
> >>> done within TeXlive, or by making the MikTeX installer work with a
> >>> preinstalled TeX system.
> >>>
> >> I agree. Having rpm's is good just for a group of Redhat based
> >> systems.
> >
> > Yes and no.  The most important thing would probably have a _file
> > list_, or package list.  Whether one generates RPMs or DEBs or
> > whatever else from such a list would then be a secondary
> > consideration.
>
> As someone who has to support actual (500+) *users* on a mix of
> platforms in which RH-based systems (eg FC) play an increasingly large
> role, can I make my 2¢ pitch for a TeXLive RPM?
>
> Thomas did a brilliant job in producing and maintaining teTeX, but for
> many years the tetex{...}.rpm collection was sadly out of date. I've
> never actually found out why, or who was supplying it to the Red Hat and
> FC repos, but supporting it meant explaining to users how to download
> updated copies of the assorted bits and pieces necessary to make $x$
> work (where $x$ was the particular package, style, font, feature, or
> function the user needed for their book/thesis/essay/etc) -- because
> they had installed the default RPMs along with their OS.  These are not
> CS geeks, able to remove the RPMs and replace them with tetex from
> source, or even from binaries, but engineers, mathematicians,
> historians, linguists, physicists, and others who simply want packaged
> installations to be updatable (eg via yum).

Vendors are starting to realize that TeX is important, in part due to
very popular "mission critical" systems such as the R stats package
that use LaTeX, but also need to be conservative about making changes.
 It is worth trying to work "with" vendors by filing bugs with them
and thinking about how TL can coexist with vendor packaging.  On
debian sid you can choose either tetex or texlive and most things will
work with either choice, but I think it is fragile and confusing for
users.

> My recent tack has been to pull the tetex RPMs and replace with TeX Live
> from whatever the then current CD/DVD was. This has fixed most of the
> out-of-datedness problems, but introduced its own, because the default
> installation directory changes every year. *I* can deal with this, but
> the average user neither knows nor cares whether it goes in /usr/TeX,
> /usr/local/texlive/yyyy or wherever: they simply want it to work, and
> to be upgradable year by year. 99% of these are single-user systems,
> so the need to consider multiple users is not relevant (the few multi-
> user systems are capably handled by departmental admins who *do* know
> where to put and find stuff).

In my experience, users are very reluctant to give up the vendor
supplied packages.
They will use TL when it is installed (in a separate tree) by an
admin, but would balk at
installing it themselves or anything that meant deleting the vendor files.

> If we can settle on a stable, reliable, and predictable location where
> TL-yyyy can be installed from an RPM, with a user-writable texmf-local
> ready-configured into texmf.cnf and all binaries ready-linked to
> /usr/bin, and a message to the installer to run texconfig immediately,
> then we might be in a position to argue the replacement of the existing
> tetex RPMs with TL RPMs.

By the time we get all that done, the majority of systems will have
decent support
for virtual machines, so maybe all you need is one linux distro that has TL.

> I realise there are political arguments in the Debian field, and a
> strong feeling in some quarters that we should only provide support to
> sysadmins rather than end-users, but I worry that unless we admit that
> the move is towards single-user workstations rather than shared lab
> or departmental systems, we will end up making it so hard for users
> to switch to TeX that we will be back in the pre-tetex days.

No, for the majority of single user workstations users will live with
what vendors provide.

> None of which should be read as detracting from the fundamentally great
> job the tetex and TL communities have done in harmonising the whole
> caboodle. Let's not let this opportunity slip.

For now the priority should be to do a really good job with TL2006 so
people looking for an alternative to teTeX aren't disappointed.  It is
important to remember that TL2006
tries to support Win32, and that MikTeX and W32tex both have something to offer.


-- 
George N. White III <aa056 at chebucto.ns.ca>
Head of St. Margarets Bay, Nova Scotia