Re: Bug#176506: Proposal seconded...though very late..:-)
Wouter Verhelst wrote:
There were two reasons why I didn't change to debconf initially. Firstly
almost all real installations of exim involve a certain amount of
customisation of the config files, making debconf far less useful as
you can't re-run debconf to change your configuration without losing the
user's customisation, and secondly some of the results of running
eximconfig are more complicated than just setting variables depending on
the user's responses.
Op do 10-07-2003, om 18:21 schreef Christian Perrier:
(if people are aware of other prompting-user-but-not-using-debconf
packages, please let me know)
exim, obviously :-)
These are, of course, not good reasons. For one thing I gradually came
to realise, from looking at other packages that used debconf, that I had
slightly misunderstood how it worked; the initial description of it on
the mailing lists when it came out implied it was much more limited than
it in fact is. The other thing that happened is that Andreas Metzler
came up with a simple - and obvious in retrospect, but I hadn't thought
of it - scheme to merge the users' changes with debconf answers. I wish
I had changed to using debconf a couple of years ago.
The excellent exim4 packages that Andreas and Marc Haber (and lots of
other people) have put together do use debconf for all prompts. I could
adopt a similar system for the exim 3 packages, but I would prefer not
to because of the complexity of migrating people's existing setups. I
believe all new installations should use exim4, and so my main priority
for the exim 3 package is to make as few changes as possible to be sure
of not breaking anything for anyone who still uses it (not just because
I'm lazy, although I wouldn't blame anyone for thinking that).
I believe exim4 should be the default mailer by the time sarge is
released, at least if its maintainers believe it is stable enough (I'm
now using it myself on my server, and I believe that it is). I think
exim 3 should remain in the distribution, mainly for people with
existing installations of it. As it would not be something people making
new installations would install, I do not think there is a great need
for it to be debconfised.
If, following this policy proposal (which I generally support) someone
filed a non-wishlist bug against exim, which they would be quite
entitled to do, I would probably ignore it, though I would discuss it on
debian-devel first and if people overwhelmingly thought I ought to
change the packages then I would.