New Beta 4.58.6 released

[Glenn Steen]

On 29/01/07, Gerard Seibert <gerard at seibercom.net> wrote:
> On Monday January 29, 2007 at 12:48:22 (PM) Julian Field wrote:
>
> [snip]
>
> > Since when did his software come with a warranty? Ooohh, can I sue him
> > please? Please?
>
> I assume he is referring to the implied warranty; i.e., the product
> works as is. If you modify or attempt to use it in a way not expressly
> approved by the author, you have voided the warranty.
>
> This who matte seems rather strange though. The following two post were
> just placed on the Postfix forum.
>
> // QUOTE //
>
> In the past, people did experience MailScanner corruption problems.
> This corruption was traced back to queue files being moved between
> Postfix queues on different file systems. This is just one example
> of what happens when software approaches Postfix internals via
> inappropriate methods.
>
> It does not matter if this particular problem has been fixed.  It
> should never have been allowed to exist. The MailScanner authors
> should have used a documented EXTERNAL interface, and if no suitable
> interface existed, then they should have proposed one - for example,
> provide a patch for adoption into the MTA.
>
> Case in point: there is no EXTERNAL interface to find out when a
> file in the "hold" queue is ready for access; I have not been asked
> to provide such an interface, nor have I seen a proposal to provide
> support for such an interface. Absent a supported interface, it is
> very well possible that MailScanner moves queue files around at an
> inappropriate time; if not today, then some time when I revise the
> Postfix queue internals. And then people will lose mail.
>
> As long as the MailScanner people ignore common software engineering
> practices, such as the use of documented EXTERNAL interfaces, they
> will put the mail of their Postfix users at risk. The statement
> that some operator never saw a problem is meaningless.
>
>         Wietse
>
> // END QUOTE //
>
> // QUOTE //
>
> It seems to me that all that is needed is for Mailscanner to propose an
> external API that Postfix could implement to allow Mailscanner to access
> the information they want and do the processing they want without
> grovelling through undocumented queue files.
>
> Paul Tomblin
>
> // END QUOTE //
>
> It has occurred to me that since Dovecot wrote an API to Postfix that
> Mailscanner should be able to accomplish the feat also. It would appear
> to me at least, that since Postfix is the 'Parent Program' and
> 'Mailscanner' is attempting to use the services that it offers, that
> Mailscanner should take the lead in developing an interface that meets
> with the Postfix developer's satisfaction.
>
Hi Gerard,

If memory serves me, this type of solution has been explored a bit in
the past, with someone (don't remember who) using one of the
"documented external interfaces" to grab the messages, take them out
of the loop (by writing them to files) that were different from the
queue files only in organization, not much else, letting MailScanenr
use the sendmail convenience command to reinject the mails. At least
that is what my decidedly flakey (ATM:-) memory tells me.
Never met much approval in any camp.

But having said that, it might not be an absolutely unrealistic view
on the matter...

Then again, all we really would need is some prior notice (and perhaps
some explanation:-) to keep doing things as we do now.

-- 
-- Glenn
email: glenn < dot > steen < at > gmail < dot > com
work: glenn < dot > steen < at > ap1 < dot > se