[tex-live] TeXLive 7 with cygwin

[Fabrice Popineau]

* Jeremy Gibbons <Jeremy.Gibbons@comlab.ox.ac.uk> writes:

> Cygwin does indeed come with its own tex, but I don't think that was
> the issue. The TeXlive installer installed at least two texmf.cnf
> files, one in texmf/web2c and one in texmf-var/web2c. They are
> wildly different (at least, diff decided the simplest explanation of
> their differences was to quote the whole files).

Probably  because one of them  was in binary  mode while the other one
was  in  text mode. Only   the first  few variables  (TEXMFMAIN, etc.) 
should be  different. The problem of working  with the texmf/  tree is
that you can't upgrade  easily. Any modification will  be lost  in the
future. Setting a copy of the config files in the VARTEXMF tree is the
easiest way to ensure your modifications will  be kept. Also, by using
a dedicated  texmf.cnf file in VARTEXMF,  I can make sure your various
texmf  trees (TEXMFMAIN, TEXMFLOCAL,   VARTEXMF)  point to  the  right
directories.


> But it occurs to me now that when I was trying things out, I wasn't
> aware that cygwin had installed its own tex. I may indeed have
> jumped to the wrong conclusion.  Would it be useful to you if I
> tried to investigate the problem further, or are you not interested
> in supporting tetex with cygwin?

I can't support tetex with cygwin mainly because tetex should compile
out of the box under cygwin, or at least, such is the goal of
cygwin. The main problems I see with cygwin :
- complex to administrate (path, quoting, drives), not Windows
oriented and very unfriendly to Windows users
- much more weird, it tries to emulate Unix calls, but sometimes at
the price of slowness; it is the case especially for the stat() call, 
which is heavily used by kpathsea. 

This latter reason alone would refrain me to use cygwin and make me
prefer to switch to some native version of the programs.

Fabrice