Re: texconfig in tetex-3.0 again

On 2005-11-03, C.M. Connelly wrote:
> "GW" == George White <[email protected]>
> 
>     GW> 1.  for security reasons, many sites multi-user systems no
>     GW> longer allow users to write to shared areas (e.g. /tmp,
>     GW> etc.)
> 
> Huh?  Restricting access to files that might actually be read and
> used by multiple users (like generated font files) makes some
> amount of sense.  But restricting /tmp would be (1) insane,
> and (2) break huge numbers of programs that depend on being able
> to write to /tmp (or, more securely, to some subdirectory of
> /tmp).

Indeed.

>     GW> 2.  as a consequence, and because disk is cheaper than sys
>     GW> admins, many packages have moved to a configuration that
>     GW> is installed and configured by each user on a personal
>     GW> workstation or entirely within the user's "home" area on a
>     GW> multi-user system.  
> 
> You must be kidding me.

Again, indeed.

>     GW> Given this background, administrators configuring teTeX
>     GW> for multiple users should realize they are now an atypical
>     GW> cases and should expect "surprises" and be prepared to do
>     GW> a little extra work, including setting up a special
>     GW> tetex-admin account, changes to the default texmf.cnf,
>     GW> patching or renaming texconfig, etc.
> 
> Sorry, George, but I think you're on crack here.  Yes, many users
> have their own installations of TeX on their PC, laptop, or what
> have you.  But teTeX qua teTeX (i.e., not repackaged ala TeX Live
> or Gerben Wierda's iInstaller packages) is usually going to be
> installed on *nix machines, which are multiuser by default.

Absolutely.

> The commands should have been left the way they were -- as
> system-level commands, with new commands with new names added for
> bold and experienced users to run to handle their own
> configurations.
> 
> By changing the names of the commands run at the system level, you
> not only mess with the heads of experienced sysadmins, but you
> also potentially break any scripts that might have relied on those
> names (and their functionality) remaining consistent.

That was an astonishingly bad decision.  I have tons of
software, much of it written more than a decade ago, which has
happily been using TeX to produce all kinds of printed output
for people who have never heard of TeX and who would be quite
surprised to find that it had been doing things for them.  Now,
to continue to provide these services, I have to resort to
manually constructing my own TeX distribution or else waste time
wading through all the stuff that relied on it.  And what would
be the point of that in a world where people think it's fair to
just capriciously break all existing uses on a whim?

> Keeping the
> user interface consistent and backwards-compatible is generally
> considered to be a good thing; you would think that would be
> especially true for a TeX system given Knuth's insistence on newer
> versions or potential replacements for TeX to produce the same
> output as his original program, and the LaTeX team's commitment
> to backwards compatibility for documents.

Well said.

Greg