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

Re: Bug#150514: Uer maodifications _must_ bre preserved, even is a co-admin said otherwise a few releases ago

On Tue, Jun 25, 2002 at 11:57:29AM -0500, Manoj Srivastava wrote:
>  Gordon> This is not sensible as a procedure, and I will not go out of
>  Gordon> my way to support uncontrolled system administration where
>  Gordon> the administrators do not know what they are doing and do not
>  Gordon> talk to each other.
> 	Of all the silly strawmen I have seen, this is the
>  silliest. The point I was making was that in real world environments,
>  when you have a few harried administrators, under heavy load, and too
>  many machines to look after, perfect communication, and even perfect
>  memory of what one did a few months ago on some machine is something
>  that may not be reasonable to expect. You may live in an idealized
>  world, but I suggest that Debian may not always be used in  a place
>  called perfect. 

This is not what I read from your email, but I am happy to believe that this
is what I should have read. Instead I read you wanted to support people
who said "A" and then went about their business as if they had said "B"
(which is difficult to support).  I still stand by my comment that, no matter
how busy people are, packages should not be blamed for admins making
changes which
do not follow the instructions for making changes supplied with the package.
I noticed that this has been backed up by a few other commentators. 

> 	Users seem not to mind when dpkg asks them about conffiles
>  from time to time, so where are you drawing this conclusion from?

>From tons of bug reports which were filed the last time people thought I
was asking too many questions... In addition, potentially I would have to
ask questions hundreds of times a week, from a background process which
has no interface, to a user who may not be logged in (ie when modules
are loaded/unloaded by kerneld or similar). Personally I do not use the
package in this way, but past experience has taught me that if it is possible
to use something in a particular way, then Debian will have someone doing it.
Finally, I would mind if a package continuously asked me the same question
over and over again. Look at "rm -i", the one time you should have said "n"
is the time you are in automatic and entered "y"... I guess the reason why
-i was voted the worst addition to Unix ever *.

* I made this up, but it is probably right.

Anyway, somewhere else in this thread hierarchy you can find the proposal for
setserial. I hope this is sufficient to fix your Policy issues.

At the very least this discussion has shown that setserial should not be
blamed for update-modules, which was the original issue in your bug report.

>  Indeed, I like the fact that sgpkg does ask me every time, and gives
>  me the option to see a diff and choose, based on the changes made,
>  whter or not to accept *this particular change*. 
> 	I think that for configuration file handling dpkg has set the
>  gold standard that we ought to be emulating.

Equally I find it confusing that dpkg alerts me to between 1 and 20
conf file changes every time I do an update (not that often). I usually
only know the function of some of these, so the comfort value of the
questions is lost as I do not really understand if it should be y or n. In
this I refer to conf files I have never edited. Hmmmm, now I think about it
these confs are probably in breach of Policy for changing themselves
automatically and being conffiles... a package printinghelp in this
matter during preconf does not help as it flies past among the other 200

>  The more you sweat in peace, the less you bleed in war.

At last, something we agree on from the start. :-)

Gordon Russell
PGP Public Key - http://www.dcs.napier.ac.uk/~gor/pgpkey.txt

To UNSUBSCRIBE, email to debian-devel-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

Reply to: