v4.85.1.0 Beta for Linux RPM Available

Glenn Steen glenn.steen at gmail.com
Mon Feb 23 12:46:25 GMT 2015


Yeah well, the core parts of the postfix code is all JF, all I did was
(at the time when the p-records were introduced to the queue file
structure) hack up some not too inefficient code to make it cope with
the new stuff. Had to delve deep into the innarts of both postfix and
MailScanner, but ... Theat was a few years back;-).

I don't know of any development in postfix that would radically change
the frontline between MS and PF... They want us to sacrifice
efficiency to protect against possible corruption by using "published
interfaces" (LMTP/SMTP would make it wellnigh impossible to keep the
"batch scanning" functionality). We want to keep doing things as
efficiently as possible.

A fun fact is that during one of Jules rows with Wietse, Wietse
actually detailed all that needed be done (as a dettering example,
mind you) to "steal" the messages from the queue, break it apart,
reconstruct a new message (pretty much identical, but with a
completely new queue file/ID). Jules only comment was "Yes, that is
exactly what we do". After that, the lines of communication were a
bit... strained:-).

Anyway, unless there is something significantly different that can be
done vis-a-vis how postfix can interract with MailScanner, there
simply is nothing to be done. At least nothing that will improve
MailScanner. And mind you, I've been using Postfix/MailScanner
successfully for more than 10 years now... Deprecated indeed;-)

Cheers!
-- 
-- Glenn

On 19 February 2015 at 00:23, Rick Cooper <rcooper at dwford.com> wrote:
> Jerry Benton wrote:
>>> On Feb 18, 2015, at 11:18 AM, Rick Cooper <rcooper at dwford.com> wrote:
>>>
>>> This applies to SpamAssassin as well. I sent Jules code to move
>>> MailScanner to use the Spamd process years ago. Now the clamav
>>> module had a lot of issues because when internal defs within the
>>> clam code changed the module would puke until someone patched the
>>> module but it makes no sense to me why anything MailScanner shares
>>> should be based on a perl module if there is a daemon available to
>>> communicate with. Spamd protocol is pretty simple and fairly easy to
>>> integrate within MailScanner. The difference in speed is pretty much
>>> nil but the difference in MailScanner memory usage per child is
>>> significant. When I originally worked with JF to integrate the clamd
>>> code his biggest hesitation was he wasn't comfortable with network
>>> code... I would bet that was the biggest reason for handling postfix
>>> the way it's handled as well. Whoever is currently developing MS
>>> should really look at moving toward spamd support in place of the
>>> perl module.
>>
>>
>> Rick,
>>
>> Please send me the working code you have for this. I will add it to
>> the development of the next version. The same is true for Postfix
>> handling if you have anything. I have spoken to Wietse Venema
>> regarding MailScanner integration of Postfix and he doesn't like how
>> it is currently done and posted a recommendation years ago on
>> postfix.org that Postfix shouldn't be used with MailScanner. In
>> short, MailScanner should be using a milter for Postfix.
>>
>> -
>> Jerry Benton
>> www.mailborder.com
>
> I will have to see what needs patch, I have not really maintained the
> patches because of my wife's health issues so it's probably a few versions
> back at least. I think the last patch I did was around the time that JF
> moved away from the project. As far as postfix, I do not nor never will use
> it. The author seems pretty full of himself without real justification. I
> just remember the issues and it seemed they were all related to how postfix
> handled it's mail queue and I believe the proper way discussed at that time
> was it should be redelivered through some process which would have required
> sockets again. You know, I think Glenn did *a lot* of the postfix code and
> seems to know it very well. Dunno if he has time to take a crack or not.
>
> Rick
> --
> MailScanner mailing list
> mailscanner at lists.mailscanner.info
> http://lists.mailscanner.info/mailman/listinfo/mailscanner
>
> Before posting, read http://wiki.mailscanner.info/posting
>
> Support MailScanner development - buy the book off the website!



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


More information about the MailScanner mailing list