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

Re: Bug#224828: Split config file is worrying...

Andreas Metzler wrote:

| On Wed, Jan 07, 2004 at 07:33:41PM +0800, Cameron Patrick wrote:
| > Could you avoid the debconf question and use a similar mechanism to what
| > the current XFree packages do: generate the file from conf.d, unless its
| > md5sum has changed from the generated version.  That way you don't need
| > to have a debconf question, and when someone edits their exim4.conf
| > file, exim Does The Right Thing and doesn't scribble all over it.
| I don't want to dump conf.d, I want to keep it around.

Yeah.  But you were proposing a debconf question to allow users to
choose between conf.d and a monolithic config file; my suggestion was
that the debconf question was unnecessary and that a better way of
finding out whether or not the user wants to use a monolithic
configuration by seeing if they've edited said config file by hand.

The logic would look something like:

- have an /etc/exim4/exim4.conf.template as you described
- compare the md5sum of /etc/exim4/exim4.conf.template with an md5sum of
  the last automatically generated version.
- If they're different, copy the admin's exim4.conf.template to
  /var/lib/exim4 rather than looking in conf.d.  The admin could perhaps
  force the use of conf.d by running update-exim4.conf, if that was
  ever necessary later.  Don't update the stored md5sum.
- Otherwise: update-exim4.conf should generate the config file in
  /var/lib as it does now, and also generate an /etc/exim4.conf.template
  (with comments kept in it) so that if the admin ever wants to make
  wholesale changes and throw away conf.d, this can be done easily using
  the current configuration as a base.  Update the stored md5sum.

Thus the admin gets to choose between using conf.d and a hand-edited
exim4.conf.template as you described in your previous post, without
having to face Yet Another Sodding Debconf Question about it.

Apologies if the above was obvious to you the first time and I'm still
missing the point somewhere.


Reply to: