Bug#1119165: exim4: Depends on db5.3

[Andreas Metzler]

On 2025-11-16 Marc Haber <mh+debian-packages at zugschlus.de> wrote:
> On Sun, Nov 16, 2025 at 08:09:17AM +0100, Paul Gevers wrote:
> > On Thu, 13 Nov 2025 15:28:59 +0100 Andreas Metzler <ametzler at bebt.de> wrote:
> > > I have switched to sqlite in 4.99~RC3-1, and the change is included in
> > > 4.99-4 which was uploaded to unstable today.

> > I saw apt-listchanges pop up this morning while upgrading my forky
> > system. As exim4 doesn't migrate things for the user, a big fat warning
> > in the release notes seems appropriate? Or did I miss-judged the impact
> > for the user? If so, please clone and reassign this bug to
> > release-notes.

> The default configuration doesn't use the db backend, but we should
> definetely issue a warning that advanced users using that feature will have
> to move.
[...]

Hello,
I had added NEWS entry which should serve as warning:
|   Use sqlite instead of berkely db 5.3 for exim hints databases and dbm
|   lookups. This can require changes that cannot be handled by the exim4
|   packages' maintainer scripts:
|   + Exim hints databases will be automatically removed on upgrades.
|   + Databases accessed with DBM loookups (dbm, dbmjz, dbmnz) will need to be
|     rebuilt in sqlite instead of berkeley db format.
|   + Vacation-like databases ("once" autoreply transport option, "once"
|     and "vacation" exim filter commands) will also need to be removed.

Comments/corrections are welcome.

When bdb was still free (i.e. up to 2014) we used to have this kind of
upgrade on a regular basis without doing anything special except a note
in debian/changelog. Still, with bdb-upgrades not happening anymore users
might have lost the knowledge to deal with them.

FWIW I have just doublechecked what happens with
8X---------------
testit at argenau:~$ head .forward
# Exim filter
vacation
testit at argenau:~$ file .vacation
.vacation: Berkeley DB (Hash, version 9, little-endian)
8X---------------

The respective incoming message is defered:
2025-11-16 13:45:19 1vKc87-00000000CNd-1jbf <= ametzler at bebt.de H=localhost (argenau.bebt.de) [::1] P=esmtp S=480 id=20251116134519.047592 at argenau.bebt.de
2025-11-16 13:45:19 1vKc87-00000000CNd-1jbf => testit <testit at localhost> R=local_user T=mail_spool
2025-11-16 13:45:19 1vKc87-00000000CNd-1jbf == >ametzler at bebt.de <testit at localhost> R=userforward T=address_reply defer (2): No such file or directory: Failed to open sqlite3 file .vacation when sending message from address_reply transport: No such file or directory

So, yes the advice in NEWS is correct, but at least there is no silent
breakage.

I will wait a bit (perhaps there is something I missed) before cloning to
release-notes.

cu Andreas
-- 
`What a good friend you are to him, Dr. Maturin. His other friends are
so grateful to you.'
`I sew his ears on from time to time, sure'