Re: Messages from the ietf list for the week ending Sun Aug 16 14:36:38 2020

[tom petch <[e-mail]>]

On 17/08/2020 10:43, tom petch wrote:
> On 16/08/2020 19:58, Andrew G. Malis wrote:
>> On Sun, Aug 16, 2020 at 2:37 PM John Levine <[e-mail]> wrote:
>>>
>>>    1 |   3014 | Wes Hardaker <[e-mail]>
>>
>> Is it HTML formatting, bottom quoting, or email header explosion (or some
>> combination of the above) that leads the smallest individual email on the
>> list this past week to be 3014 bytes in length? (I didn't bother to
>> average
>> the email lengths from multi-posters).
>>
>> For what it's worth, I'm trying to send this without formatting and with
>> minimal quoting to see how small I can get it. Now I have to try to
>> refrain
>> from posting anything else on the list this week ... :-)
>
> Andrew,
>
> e-mail header explosion by the ESP in part, HTML in part.
>
> The plain text announcement of an RFC on the announce list is 2kbyte to
> 3kbyte.  In 2012, that was up to 6kbyte by the time I received it.  In
> 2013 that jumped to 9kbyte and in 2015 to 12kbyte and in 2016 to
> 16kbyte.  In 2019, it went down to 11kbyte.  The difference is mostly in
> the x- headers with X-Microsoft-Antispam-Message-Info
> being enormous.
>
> HTML varies a lot.  Your message reached me as 13423byte; removing HTML
> takes that down to 12193byte while removing the X- headers cuts that to
> 6207byte.  At the MSA stage it was probably 1270byte if it were plain
> text only.
>
> BUT this message is a very conservative use of HTML.  Some MUA like to
> include the Microsoft style sheet which can expand a message from
> 10kbyte to over 100kbyte - yes, one hundred. As a rule of thumb I would
> say that a multipart alternative with HTML is about twice the size of a
> plain text message.
>
> I select the MUA option of plain text and think that the security of the
> IETF system, and the usability of web mail, would be improved if that
> were mandatory:-)

Just to add to my earlier figures what may, or may not, be obvious, 
John's figures come from IMAP with the data on the server whereas mine 
come after the e-mail has been downloaded with POP3, which means that it 
has passed through more MTA, every one of which adds a header or two, so 
John records me as posting one e-mail of 11404 byte but by the time I 
had downloaded it, that had grown to 18646 byte.  I think that John's 
figures include all the headers that it took to get the e-mail to the 
server.  I note that the addition of large anti-spam headers can occur 
prior to arrival on the server, in which case they would count toward 
the IMAP figure, during the download with POP3 or both; it depends on 
the policy of the ESP involved.

If, by  bottom posting, you mean keeping all the previous exchanges, 
then each addition is likely to be a few kbyte so it builds up and a 
single, plain text e-mail can grow to tens of kbyte, even 100 or more. 
Where the text goes, in plain text, has little or no impact.  Where my 
response includes the e-mail headers from previous posts, then I tend to 
edit the headers to just sender and date, which saves a few bytes but 
saves several lines of display so users have less to scroll down.

> HTH
>
> Tom Petch
>>
>> Cheers,
>> Andy
>>
>
> .
>