239797 - Delete from POP3 server action on partial message.

Status: RESOLVED FIXED

(MailNews Core :: Filters, enhancement)

Description

[Serguei Trouchelle]

User-Agent:       User-Agent: Mozilla/5.0 (OS/2; U; OS/2 5.0; en-US; rv:1.6) Gecko/20040113 TorMozilla
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.6) Gecko/20040113

When getting messages from POP3 server, while filter is based on subject or
other header parts, Mozilla always download whole message before deleting it.
There is a lot of garbage that has to be downloaded from servers, even when
filter action can be run after downloading ten header lines, and save bandwidth
and traffic. It would be very great to make Mozilla able to delete it asap, just
like Outlook, that already has this functionality. 

Reproducible: Always
Steps to Reproduce:

Comment 1

[Mike Cowperthwaite]

POP3 servers don't reliably support the TOP command which allows downloading of 
just headers; that command is described as "optional" in RFC 1939.  That doesn't 
necessarily mean it couldn't be *attempted* in Mozilla...

I did find a little discussion on the Web which implied that TOP is not only 
optional, but that it's poorly defined and may exhibit inconsistent behaviors -- 
specifically whether or not issuing TOP marks a message as read on the server.

IMAP normally sends just headers until a body download is requested, but the 
current junk controls require analysing the entire body; a request to perform 
junk analysis only on headers (for IMAP) is bug 215941.

Comment 2

[Christian Eyrich]

Additionally issuing TOP, getting the headers and then issuing RETR would slow
down message retrieval for all customers.

Comment 3

[Mike Cowperthwaite]

See also bug 185184.

Comment 4

[Howard Chu]

(In reply to comment #2)
> Additionally issuing TOP, getting the headers and then issuing RETR would slow
> down message retrieval for all customers.

I typically receive 200 emails a day of which at least 75% are spam; on bad days
the ratio goes up to 95% spam. I don't think my spam-to-content ratio is very
unusual. In my case, a separate TOP/header pass, filter/junk removal, finally
followed by RETR (only retrieving the 5-25% that are real) is actually faster
than blindly downloading everything, especially if I'm traveling and away from
any broadband connections.

I find this approach to be extremely helpful, which is why I submitted a patch
for this on bug 185184.

Comment 5

[Mike Cowperthwaite]

This has actually been fixed for some time due to Howard's work on bug 185184.