Clarification of Policy and Packaging manuals requested

[Manoj Srivastava <srivasta@datasync.com>]

Hi,

	I would like to request a discussion about Configuration files
 and the conffiles mechanism. We knwo the following:

 a) Any configuration file required or created by a package has to
    reside in /etc (POLICY 3.3.7)
 b) Most configuration files under /etc [where else could they be?]
    are also supposed to be conffiles (POLICY 3.3.7). So we know that
    there is a gret deal of overlap in the two categories, [namely,
    Configuration files and connfiles]
 c) conffiles are meant to be user modifiable files, and the packaging
    mechanism goes out of its way to make sure user mods are not
    lost. (PACKAGING 9).
 d) The conffiles should be specified as absolute pathnames (PACKAGING
    9.1).

	It was my understanding (not shared by at least a few other
 people) that the conffiles were a propoer subset of the Configuration
 files set, and hence by rule (a) were constrained to be in /etc too.

	It has been proposed that that is not the case, that conffiles
 are an independent classification.

	If that is the case, then under one interpretation (a) becomes
 meaningless, since it only applies to the small subset of files that
 are configuration files but are not conffiles (if indeed conffiles
 are unconstrained).

	I would like clarification of the criteria by which one may
 classify a file a conffile (I hope we are all in agreement about the
 distinction between configuration files and program data files?), and
 whether conffiles are indeed unconstrained by (a) above. If the
 latter is true, lacking any criteria for selection of conffiles, (a)
 above is meaningless.If indeed my interpretation of the manuals is in
 error, I would like these clarification to be put into the policy
 manuals. I consider this ambiguity a defect in the manuals.

	In my opinion, I would ask for conffiles to be exactly the set
 of configuration files, plus exceptions decided by consensus no the
 mailing lists (lmbdacore and dosemu initial files have been proposed
 as conffiles; and the reason for the extention cited as being too
 large to be candidates for the root file system).

	I would appreciate if the authors of the manuals in question
 also tender an input, hence I'm taking the liberty of copying this
 message to them directly.

	manoj
-- 
 "...and it's finished!  It only has to be written." Karl Lehenbauer
Manoj Srivastava  <srivasta@acm.org> <http://www.datasync.com/%7Esrivasta/>
Key C7261095 fingerprint = CB D9 F4 12 68 07 E4 05  CC 2D 27 12 1D F5 E8 6E