Quote

[Antony Stone]

On Saturday 30 August 2003 8:19 pm, Chris Trudeau-Personal wrote:

> > Ouch. Wietse Venema is the author of postfix, so I'd regard his
> > opinions about anything postfix-related to be pretty relevant.
>
> I would too, except that he added the comment about all other MTAs which in
> my mind indicates a prejudice...against the concept of how MailScanner is
> architect.
>
> I respect his comments and opinions, although I have had zero problems with
> the postifx/mailscanner system I'm running.  The reasoning that Wietse uses
> is based on the guarantee of queue file uniqueness and the possibility that
> mailscanner could break that guarantee...I'm surely not as smart as
> everyone on this list, so what are the general thoughts?

Personally, having first got used to sendmail (and still happily using it),
I've always found it strange since I first learned that Postfix doesn't have
the same sort of concept for queue filename uniqueness as sendmail does
(where the name is based on timestamp with 1 second resolution and 60 year
wraparound, combined with process ID of the sendmail instance processing the
mail).

I'm not saying that's the only way to do it, but Postfix's inability to
guarantee uniqueness of filenames even for files stored on two different
filesystems on the same machine, for example, strikes me as a bit of a
design deficiency.

Antony.

--

There are only 10 types of people in the world:
those who understand binary notation,
and those who don't.