[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: /etc/sysconf ? /etc/<somethingelse> ?

>>>>> "IJ" == Ian Jackson <ian@chiark.chu.cam.ac.uk> writes:

IJ> Many packages are finding themselves in need of a directory to place
IJ> one or two small files which are read by initialisation scripts.

IJ> For example, /etc/papersize could be here (though perhaps we don't
IJ> want to move it now), as could files telling the rc.boot scripts which
IJ> keyboard map to load.  The timezone package could put its record of
IJ> which timezone has been copied to /etc here.

What exactly is supposed to go into that directory? I see a point for
cross-package system defaults like /etc/papersize. But under which
circumstances would a package be allowed to put private information
into this directory? After all, each package contains the default
configuration files. And things which cannot be defaulted in a
reasonable way get setup interactively. These configuration files
should be the record of what was the previous version of the

Using the timezone package as an example, I can obtain the
configuration information either by reading the link /etc/localtime or
by looking at /etc/timezone. Why should there be another file with the
same information? How do you keep an sysadmin from changing the files
/etc/timezone and /etc/localtime without updating the new default

If the configuration files are scripts - like those in /etc/init.d - I
see the point to separate the selection of defaults from the actual
code. And a comment in the script can tell the user how to change the

In any case: would there be some documentation of legitimate use and
contents of the default configuration files? E.g., can I put letter,
Letter, US letter, A4, DINA4, DIN a4, ... in /etc/papersize and have
them recognized by all packages which read /etc/papersize? What I want
to say is that shared configuration files only help if their use is
defined somewhere - basically like the list of virtual packages or in
an appendix to the developer guidelines.

IJ> It's clear to me that we need this directory; it's less clear to me
IJ> what it is called.

I have no strong feelings about the name of the directory
though. Previous art, like the SYSV choice, might serve as a guide,

   Dr. Lukas Nellen                 | Email: lukas@teorica0.ifisicacu.unam.mx
   Depto. de Fisica Teorica, IFUNAM |
   Apdo. Postal 20-364              | Tel.:  +52 5 622 5014 ext. 218
   01000 Mexico D.F., MEXICO        | Fax:   +52 5 622 5015

Reply to: