Sloppy error checking in MS code

[Glenn Steen]

On 17/12/06, Dhawal Doshy <dhawal at netmagicsolutions.com> wrote:
>
> Quoting Glenn Steen <glenn.steen at gmail.com>:
>
> > On 17/12/06, Mike Jakubik <mikej at rogers.com> wrote:
> >> Rob Freeman wrote:
> >>> If any of those problems happen, u have more troubles then MailScanner.
> >>> Personally, it is not MailScanners job to monitor a server for disk space,
> >>> permissions, or any other OS / HW problems.  And if someone is tampering
> >>> with permissions, well one has much more issues then getting mail at that
> >>> point.
> >>>
> >>
> >> I never stated that MS should monitor my system, it should however fail
> >> gracefully and not eat messages.
> > Quite a resonable standpoint. And we're all pretty sure that this is
> > the state of things too;). But we should look, yes.
>
> Time for me to add some facts (of course for my point of view) here,
> since i follow the postfix list very regularly..
>
> Wietse and Viktor from my little interaction with them have no problem
> with any Filtering tool (including mailscanner) as long as it uses one
> of the *documented* interfaces (how is not what i mention here, read
> the thread below for wietse's comment)
>
> The comment on the mailscanner code being sloppy was made by *ahem*
> 'Mark Martinec' of amavisd-new fame.. See this thread:

:-) Missed that pertinent point:-)
> http://groups.google.com/group/list.postfix.users/browse_thread/thread/d919f921151c8cb3/d1ac170a6838855f?lnk=gst&q=mailscanner&rnum=1#d1ac170a6838855f
>
> It is therefore understandable that Mark made this comment (and
> probably rightly so).. however let us look at the larger picture and
> those who can read code, ought to make an effort to 'audit' the code
> for the *so called* sloppiness just to re-assure themselves (and
> probably others).

Exactly.

>
>  From my point of view again, checking for free disk space and other
> such things is *not* mailscanner's job.. as long as it fails
> gracefully and consistently.. there shouldn't be a scope for a *maybe*
> anywhere. And why blame postfix? the locking for sendmail changed as
> well a few versions back (try and recollect the number of times people
> on this list have been using the wrong locking mechanism for
> sendmail).. something for exim changed as well breaking things for a
> while.

Still with you.

>
> Finally a few words about DJB, if you can code even half as well as
> him then you have all the right to criticize else just shut up and use
> something else. How many pieces of code from the last century work as
> they were documented.. till today???

(Playing at being more obtuse than I hopefully am:-):
Where did I criticise him? Nowhere, of course. If I were to do that,
it'd be along the lines of being a bit opinionated (not unlike
Wietse:-)... But they've earned the right to have strong opinions, so
I'm fine with that... :-):-).

The criticism in the PF users list is still mainly founded around the
_perceived_ groping of queue files. This is still where I'd
concentrate my auditing, when I have the time to do it. Notable is
that at least some of the folks are becoming more ... varied in their
approach:-).

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