Thank you, Aaron.
I am using Paul's debian pkg of 2.0.4 dated the 16th of january,
thinking it might include the patch. It appeared to install but the
alias / user functionality appears to remain as it was with the 2.0.3
dated the 12th.
Could be I need to force the apt-get into really getting the new package
- I will look into the instructions for that formerly provided by Petru
Paler.
Or do I need to apply the dsn.patch to 2.0.4... doing something like
apt-get source dbmail-pgsql, then unpacking, patching dsn.c, then in
some way recompiling? Or I can wait til tomorrow when the guys smarter
than me are available (they are sleeping.)
Thanks again to all of the dbmail developers and esp Aaron and Paul
- April
Aaron Stone wrote:
On Sat, Jan 15, 2005, April Lorenzen <[EMAIL PROTECTED]> said:
So what do we do with this:
>
> dbmail_users (id, user):
> 1 foo at server
> 2 foonew at server
>
> dbmail_aliases (alias, deliver-to):
> foo at server 2
>
> That is to say, we're trying to get all mail to 'foo at server ' to
> land in the INBOX of 'foonew at server '. What happens:
>
> Currently:
> Mail for 'foo at server ' goes to user 1.
>
> Aaron's Proposal:
> Mail for 'foo at server ' goes to user 2.
>
> Thomas' Proposal:
> Mail for 'foo at server ' goes to user 2 and user 1.
Suggested correction to "what happens in this scenario" with Aaron's
proposal:
Mail for 'foo at server' goes to user 2 and user 1 - because the admin
makes an explicit alias for any user who should receive mail addressed
to 'foo at server':
dbmail_aliases (alias, deliver-to):
foo at server 2
foo at server 1
Just as I believed when I set up my dbmail 2.x server months ago - that
I would need an alias for every user. So Aaron's proposal relieves me of
having an alias for each user - which of course I could avail myself of
right now - since the new dbmail doesn't actually require an alias
anymore... but I would definitely have no problem with the the concept
that if I am going to use ANY alias for user 1 ... I must ALSO add an
alias explicitly to user 1.
This seems to me to be the most backward compatible with the way dbmail
aliases and users used to work, while retaining the advantages of not
requiring an explicit alias if the only deliver-to for the address *is*
a user matching the address.
I just drew this out on paper and tested it, and you are absolutely
correct that this would mimic DBMail 1.x behavior, and make quite a bit of
sense as instructions; basically, aliases override usernames.
I'll post a patch in a few minutes.
Aaron
_______________________________________________
Dbmail-dev mailing list
Dbmail-dev@dbmail.org
http://twister.fastxs.net/mailman/listinfo/dbmail-dev
