Aaron Stone wrote:
But, for deleting all of a user's mail and mailboxes, I think we need -y
to make *really* sure that the person is doing what they intend. I think
this should go into 2.0 asap; Paul would you mind coding it?
No go for 2.0 I say. That's changing expected behaviour for a main cli
interface. Capital sin. And even for 2.1 I'm against it.
The need to add -y to get things actually done with dbmail-util has caused
enough headaches already. Such policy decisions should not be hard-coded. Let's
get our interfaces right first shall we.
Ok, so people can shoot themselves in their feet. I don't really mind if they
consciously chose to do so and any loss of data is in alignment with documented,
and therefor expected behaviour. Rather thus, than *not* deleting data when the
sysadmin is trying to remove mailboxes and getting an error, even though an
error condition is in effect.
Policy is what we're dealing with in situations like dsn.c where delivery is
involved. However, dbmail-users and dbmail-util are about providing powertools
to administrators. People should be able to use them in scripting frameworks
that implement the actual policy with respect to data and user management.
So I guess you hit a nerve here :-) If I spent any time on this, it is to make
sure dbmail actually behaves as expected. More test-cases to flesh out.
Btw, the bug involved was a minor typo:
Index: db.c
===================================================================
--- db.c (revision 1647)
+++ db.c (revision 1663)
@@ -1817,7 +1817,7 @@
}
for (i = 0; i < n; i++)
- message_idnrs[i] = db_get_result_u64(0, 0);
+ message_idnrs[i] = db_get_result_u64(i, 0);
db_free_result();
/* Delete messages one at a time. This means that we're
--
________________________________________________________________
Paul Stevens mailto:[EMAIL PROTECTED]
NET FACILITIES GROUP PGP: finger [EMAIL PROTECTED]
The Netherlands________________________________http://www.nfg.nl
