[Top] [All Lists]

Re: slight update to draft-macdonald-antispam-registry

2011-05-10 14:12:58

Jeff Macdonald wrote:

If you've looked at the draft, and I mean those besides Murray, what
has kept you from implementing it?

My opinion.

In general, like all things, the good ideas are the ones that solves a real problem.

The SMTP state machine is already well defined and it required reply codes to drive the client/server automation. Everything else is just "candy" like extended reply codes are optional, not required to drive the C/S protocol and no one can depend on it solely. IOW, even modern clients must also support legacy server with pure SMTP responses.

So you have to provide a good reason why it should be considered. Its a matter of technical writing style but I generally like to start with problem statement and then show why/how the proposed idea can help address the problem and maybe also add value (i.e. 2 for 1 benefit).

That is why I suggested the text about how it (specific extended codes like X.8.26) can help resolve some of the negativity behind GreyListing with advanced Retry methods, but as I mention above, for X.8.26, anyone with such an advanced retry method, it can't depend solely on this specific code. You got the original Harris recommendations of 451 and/or 4.7.1.

Finally, this is a subject of opinions. IMV, the I-D extended responses are focused for bad guys rejects, not good guys rejects except for the new X.8.26.

So in general, for my own system, I don't particular like to give bad guys "clues" why their transaction failed. For false positives, it can help the support process in a DSN, but for bad guys, its not like they will listen to them anyway or even get them or maybe redirected to someone else. They are going to blast away anyway.

So if you can extract how these "bad guy" transaction rejection responses can help good guys solve a particular problem, more readers will consider it.


Hector Santos

<Prev in Thread] Current Thread [Next in Thread>