Re: Fetching short text snippet for mails

Hi Arnt,

--On December 2, 2014 at 6:06:20 PM +0100 Arnt Gulbrandsen 
<[email protected]> wrote:

>> Efficient as in eliminating the current extra round trip of
>> doing a FETCH BODY.PEEK[<section>]<<partial>>? If so, then I
>> think "SNIPPET" should be an alias for BODY.PEEK[the first
>> significant text section]<<partial>>. i.e., the server does not
>> do anything more than determine the "the first significant text
>> section". The client still gets to use <<partial>> as it could
>> today.
>
> This I like. Simple for the server, simple for the client, a clear win
> for the client. The only disadvantage is that the client still has to do
> the heavy text extraction. But it needs that code anyway if it is to work
> with unextended servers.

The problem is I am not sure how well this works with HTML where there may 
be a lot of markup prior to any real text being present. That will force 
clients to request a larger enough number of octets in <<partial>> so that 
they have a reasonable chance of extracting "real" text. But then the size 
of the response gets a lot larger and thus it becomes less efficient 
overall - defeating the whole purpose of doing that.

Do we have a summary of client behaviors that shows the set of FETCH 
requests clients currently do to populate their message lists? I think it 
would be useful to see exactly what all the clients do - in particular for 
cases where there are a large number of new messages (e.g., loading a 
mailbox for the first time) vs just a few new messages (sync'ing new mail). 
Perhaps seeing that will clarify exactly what is inefficient and could be 
improved.

-- 
Cyrus Daboo

_______________________________________________
imapext mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/imapext