Re: [webmin-l] virus scanner

[Daniel P. <da...@ri...>]

Craig White <cra...@az...> writes:
> On Mon, 2005-09-26 at 12:29 +1000, Daniel Pittman wrote:
>     Craig White <cra...@az...> writes:
>     > On Mon, 2005-09-26 at 11:46 +1000, Daniel Pittman wrote:
>     >> Craig White <cra...@az...> writes:
>     >> > On Mon, 2005-09-26 at 10:54 +1000, Daniel Pittman wrote:
>     >> >> da...@so... writes:

[...]

>     >> Oh, I don't question that MailScanner *works*.  I just question the
>     >> wisdom of using a product that explicitly depends on undocumented,
>     >> internal features of other product to function correctly.
>     >>
>     >> When Postfix or Sendmail decide to change their internal queue format,
>     >> which has happened before and will doubtless happen again, MailScanner
>     >> will no longer function correctly.
>
>     [...]
>
>     > my understanding is that Postfix has actually implemented a permanent
>     > feature to hold and interrupt mail queue for this to operate and that is
>     > what MailScanner does.
>
>     Absolutely.  The ability to hold mail is a standard feature of Postfix.
>
>     It even has a good, solid interface.  You can hold mail based on any
>     number of standard table lookups, or using administrative tools run from
>     the postsuper shell command.
>
>     MailScanner, on the other hand, moves files between queue directories.
>     That /isn't/ a public interface: that is fiddling with the internals of
>     the mailer, and hoping that what worked yesterday will work tomorrow.

[...]

> it all comes off as sort of FUD - but as I said, I am under informed
> and cannot intelligently debate this but invited commentary from the
> MailScanner list...
>
> http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0509&L=MAILSCANNER&P=R51276&D=0&H=0&I=-3&O=T&T=0
>
> http://wiki.mailscanner.info/doku.php?id=documentation:configuration:mta:postfix:politics

I note that the wiki page you reference states exactly what I do:
Mailscanner depends explicitly on an interface to the MTA that is not
publicly supported.

The mailing list also echos my statements: if the queue format changes,
MailScanner needs to be changed to adapt to that.  I even agree with the
authors: another change is reasonably unlikely.  

> What I do know is that MailScanner is brilliant, effective, easy enough to
> set up, maintain and has worked with all of the various versions of Postfix
> (2.1.x & 2.2.x) & Sendmail that I have paired it with 

I have no disagreement with these statements.

> and I have absolutely no fear of anything breaking from updates (as if
> updates don't occasionally break things anyway).

This, however, is where I see our approach differ:  the last time I had
anything break on a server I support, due to an upgrade, was December,
1999.  

> I fail to understand what point you are actually trying to make - that
> all of your solid research means that I shouldn't be doing what I am
> doing because of what may occur in the future? 

No: the point I am trying to make is that I, personally, consider the
design of MailScanner to be a poor design, and that I, personally, would
not recommend it to anyone due to that design.

I also tried to make clear exactly what those technical problems were,
why they were problems, and provide an recommendation for an alternative
package that provides the same functionality without the same design
flaws.

I don't particularly care if you continue to use MailScanner or not, or
whatever.  That isn't of interest to me.

> From what I see on MailScanner list, spamassassin list, CentOS list,
> nahant & taroon lists, that a lot of people aren't necessarily
> troubled by the things you see in your solid research and using it
> anyway - with Postfix.

Which is, again, their prerogative.  You, and they, are welcome to form
your own opinions.

[...]

> My point - to be as succinct as I can be is that your statements -
> that MailScanner isn't appropriate to use with Postfix because of
> unanticipated changes to Postfix might break a working configuration
> is nothing but spreading unnecessary FUD.

I am a bit sad that this conversation seems to have ended up in name
calling.

I made my opinion known to a poster who asked for it.  They, perhaps,
asked on a somewhat inappropriate forum, but I figured a brief message
wouldn't hurt anyone.

I did my best to provide sound technical arguments in support of my
views, and to assist the original poster.

I don't believe that this was "spreading unnecessary FUD", and I would
rather you had stayed on firm technical ground in your arguments than
drifting into these personal attacks to shore up your arguments.

In any case, I don't believe that anyone will benefit from further
discussion, from us, on this topic.

Regards,
        Daniel