Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Address extension lowercased?

13 views
Skip to first unread message

David R Bosso

unread,
Apr 8, 2004, 1:46:19 PM4/8/04
to
Howdy,

We're using a MySQL virtual table to deliver to Cyrus IMAP via LMTP. It
looks like postfix is lowercasing the unmatched address extension (after
the +) when adding it back to the result of the lookup.

Is there any way to prevent this from happening?

We're running Postfix 2.0.19.

Thanks.

-David

Wietse Venema

unread,
Apr 8, 2004, 3:50:39 PM4/8/04
to
David R Bosso:

> Howdy,
>
> We're using a MySQL virtual table to deliver to Cyrus IMAP via LMTP. It
> looks like postfix is lowercasing the unmatched address extension (after
> the +) when adding it back to the result of the lookup.
>
> Is there any way to prevent this from happening?

Postfix case folds the input before table lookup is done.

Systems that depend on case sensitive addresses are fragile.

Wietse

David R Bosso

unread,
Apr 8, 2004, 3:58:55 PM4/8/04
to
--On Thursday, April 8, 2004 3:50 PM -0400 Wietse Venema
<wie...@porcupine.org> wrote:

> David R Bosso:
>> Howdy,
>>
>> We're using a MySQL virtual table to deliver to Cyrus IMAP via LMTP. It
>> looks like postfix is lowercasing the unmatched address extension (after
>> the +) when adding it back to the result of the lookup.
>>
>> Is there any way to prevent this from happening?
>
> Postfix case folds the input before table lookup is done.
>
> Systems that depend on case sensitive addresses are fragile.

Thanks. In our case the address extension specifies a user mailbox and the
mailbox namespace is case sensitive. The current Postfix behavior imposes
a requirement that plus addressing can only be used to deliver to all
lowercased mailboxes. This is somewhat confusing to the end user, and
therefore not desirable.

Would you accept a patch to retain case on the extension part? I'm not
sure how hard it is, but I'm asking before I put any time into it.

Alternatively an option on whether to case fold the input would also be
useful.

-David

Wietse Venema

unread,
Apr 8, 2004, 4:12:35 PM4/8/04
to
David R Bosso:

This should work for every address manipulation class that is
controlled by the propagate_unmatched_extensions configuration
parameter.

> Alternatively an option on whether to case fold the input would also be
> useful.

If case preserved extensions are introduced then certainly the
existing behavior needs to be available.

Wietse

0 new messages