Norbert Preining <[EMAIL PROTECTED]> wrote: > On Mit, 31 Jan 2007, Frank Küster wrote: >> Have you reread the discussion in bug #379089? It was closed with this >> changelog entry: > > Yes, but I didn't comment because I didn't agree. And I still do not > agree. If we do this, then we have about hundreds of config files we > should create: > dlocate \*.cfg | grep texmf-texlive | wc -l > and these are only those which are named .cfg. If one wants to go to > extrem why not make latex.ltx eg a config file.
Most of the files that end with "cfg" are not intended as configuration files. See for example Ralf's comment in #379089 about listings.cfg, or a random sample that I picked, subfig.cfg: %% This file is NOT the source for the subfig package, because almost %% all comments have been stripped from it. It is NOT the preferred form %% of the subfig package for making modifications to it. %% %% Therefore you can NOT redistribute and/or modify THIS file. ... > One might change > something in latex.ltx which changes system wide the format for quite a > lot of formats. And it would not survive an upgrade ... Yes, this is technically possible, just as I can make changes to usr/include/kpathsea/tex-file.h to change the system-wide behavior. But it doesn't make sense, whereas changing geometry.cfg does make sense. I see only one alternative to making geometry.cfg a conf(iguration)file and doing nothing, and that is to convince upstream to make dvips the default. Then we could ignore geometry.cfg. > Other example (the first I got in my hand), ebezier.sty. I change it to > have other definitions of > \newfont{\vrm}{cmr5} > \newfont{\virm}{cmr6} This is analogous to changing headers, in my view. Regards, Frank -- Dr. Frank Küster Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich Debian Developer (teTeX/TeXLive)