VeNoMouS wrote:
Sorry so this patch just stops removal of vlog records? Because you don't
actually make that very clear.

After closer reading this is a ./configure only control. I think it should probably be a command line switch. Maybe with a config option that sets if it deletes by default or not.

Its not my patch and I use cdb, so I'm asking anyone who uses it what it should do.

-----Original Message-----
From: Rick Widmer [mailto:[EMAIL PROTECTED] Sent: Friday, 24 November 2006 3:14 p.m.
To: A list for the discussion of ongoing vpopmail development.; vpopmail
Subject: [vchkpw] roamer's patches

In looking at the qmail control file sort limit problem, and roamer's
solution in particular I've found that he has provided a number of
patches that should be applied.  Two of them are related, and change the
behavior of vpopmail.

[1573275] An option to remove deleted users/domains from vlog.

In vpopmail 5.4.17, a new feature was introduced - when
a virtual user or domain is deleted, also delete the
relevant entries from the SQL log file. However, there
are cases when this might not be such a good thing, and
IMHO it should not really be the default behavior.
Various reasons to preserve history come to mind -
security audits, forensic analysis, plain old reports...

So here's a patch against the 5.4.17 source that adds a
new configure option, --enable-sql-remove-deleted, and
a couple of ifdef's around the places in the code where
the actual entries are deleted from the vlog table.
Note that this will change the out-of-the-box behavior
of vpopmail - removing the log entries is now disabled
by default - thus reverting to the pre-5.4.17 behavior
of keeping history intact.

[ 1573285 ] Fix removing log entries for deleted users in vpgsql.c

Here's a patch, from the FreeBSD port of 5.4.17, to
actually make vpgsql.c compile - the changes to make it
remove log entries for deleted users and domains had a
misspelled name of a variable and of a PostgreSQL routine.

This patch is actually meant to be applied after the
--enable-sql-remove-deleted patch (SourceForge
#1573275), but it should be trivial to fish out the
misspellings even if y'all decide not to make this
removal configurable.

Since its a fix to the pgsql module, the second one should definitely go
in.  I'm not so sure about the first since it changes program behavior.

Should I merge them both?


Reply via email to