John H. Robinson, IV wrote:
> Gus Wirth wrote:
>> I recently discovered that Google mail accounts (gmail) allows a unique 
>> form of e-mail address. It turns out you can use your e-mail name and 
>> add a plus sign (+) and some other word to form a valid e-mail address. 
>> For example, if I had an account at gmail with my user name of 
>> not-my-real-address, I could do something like this:
> 
> That has been a part of sendmail for years.
> Postfix supports +.
> qmail (by default) uses -.
> 
> I had noticed that about qmail a long time ago, but it was a halfsies
> kind of thing. I tested offsite to gmail, and gmail to gmail, and only
> one worked. I submitted a feature request / trouble call about it, and
> theyy fixed it: they removed the user+ext functionality altogether.
> 
> If it is back, I would not rely upon it always working.
> 
> What does work, and should always work, is that all .'s inside the user
> portion are ignored. u..ser, us.er, and u.s.e.r. are all the same, but
> you should be able to filter on them (untested).
> 

So, do I understand correctly: that the formally valid rfc822 addresses
(such as examples above containing a '+' or '-' separator or any number
of embedded '.'s) are handled /by convention/ by mail transport and
delivery agents in their own _special way_?

  (Where /special way/ means ignoring suffices or embedded dots)

That sounds strange -- unless there is some spec or informal but defacto
standard common agreement, how can one know what is /supposed/ to happen?

Or in other words, you only know what will happen by testing individual
pair combinations. And then I guess unwanted application /favors/ is yet
another part of the story.

If I misinterpreted, please help.

>..

Regards,
..jim


-- 
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list

Reply via email to