On Mon, 21 Apr 2003 17:31:10 +0200, Denis Barbier <[EMAIL PROTECTED]> said:
> On Mon, Apr 21, 2003 at 02:15:59AM -0500, Manoj Srivastava wrote:
> [...]
>> Over writing user changes is a violation of policy. Asking users if
>> it is ok with them if we violate policy is not good enough.
> [...]
Perhaps I should elucidate. In the cxontext of this thread, it
was obvious to me that we are talking about a one time question
whether or not is is OK to always overwrite configuration files
forever more, which causes user changes to be silently lost from that
point on.
> I would be glad to learn why ucf does it right. In your opinion, is
> proftpd a good example of configuration file handling?
The gold standard of conffile handling has been dpkg; every
time the maintainers version changed, dpkg asked if the user wanted
the changes, iff there were any user changes.
All ucf adds is a way to merge changes in maintainer versions
into the local copy, and some inchoate thoughts about characterizing
the magnitude of change in the maintainer versions (this is still in
early design phase).
Debconf and configuration files have not fundamentally changed
anything: the new maintainer version, instead of being embedded in
the package, is now generated on the fly; but the user should still
be asked if their changes are to be overwrtten, and they should be
allowed to see the differences, and perhaps defer changing their
configuration file to later manually merge in the changes.
Oh, I downloaded proftpd, and on first glance, the postinst
seems to do the right thing. Nice style too ;-)
manoj
--
((lambda (foo) (bar foo)) (baz))
Manoj Srivastava <[EMAIL PROTECTED]> <http://www.debian.org/%7Esrivasta/>
1024R/C7261095 print CB D9 F4 12 68 07 E4 05 CC 2D 27 12 1D F5 E8 6E
1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C
