[tex-live] [Fwd: Re: TeXLive Perl + latexmk (windows)]

[Manuel Pégourié-Gonnard]

Philip TAYLOR a écrit :
> That is the one point with which I would take issue.  As I have
> written elsewhere, the naive belief that all Windows users
> still live in the dark ages, have one physical drive with one
> logical partition, and that it is Good Practice [tm] to store
> programs and data on the same partition as the operating system
> is a myth that needs to be eradicated, by force if necessary !
> 
As far as I am concerned, this myth is already eradicated: even my mother does
have separate partitions for programs and data. But...

> It is, I believe, a great mistake that TeX Live, too, assumes that
> a single partition will be used to hold the whole of a TeX
> installation,

1. This is in no way specific to windows.
2. I don't think it's a mistake.

> and makes no provision for storing the binaries
> outside of the primary data hierarchy.

Well, it's hard to tell program from data in the various files from texlive. For
example, a lot of scripts do only have a stub in the "bin" dir, and the actual
program is a text file in the texmf(-dist) tree. So, from an administration
point of view, I don't see any reason to put some part of TeX Live in one place,
and some other part in another place.

And as Norbert said, the search mechanisms of kpse assume that texmf has a fixed
location relative to the binaries. But, as mentioned in a previous discussion,
the whole package infrastructure (tlmgr) does a similar assumption (even
stringer, since both texmf and texmfdist have to be fixed). So, changing this
would mean at least two *major* changes in TeX Live, and even with a very good
reason it is highly unlikely to happen.

Manuel.