[Mailman-Users] Headers and why they suck

[Eric A. Meyer]

At 14:32 +0100 4/9/02, robin szemeti wrote:
>On Monday 08 April 2002 17:38, David wrote:
>>  just think that the way they say
>>  that leaving the headers in is following standards and that there is no
>>  option to remove them is absurd especially when that is not wht he
>>  standards say.
>
>'no option to remove them' ?? what?
>
>just edit the source, how hard can it be??

    Impossible, in some cases.  Let's take an example: pair.com (a 
very large Web hosting provider, among other things) provides 
accounts with free use of "pairlist."  This is all based on Mailman 
(v2.0.6 at this point), and so far as I can tell the lists are all 
administrated through the Web-based interface.  If an option doesn't 
appear in that interface, then the list admin can't affect that 
function.  For some odd reason, pair.com doesn't want people editing 
the Mailman source that runs all its lists-- can't imagine why that 
would be.
    It's true that many accounts at pair allow for running one's own 
scripts, and that would suffice for many people.  For many others, 
who have cheaper accounts that don't permit shell access or 
customized CGI, it would be a touch more difficult.
    And as a last point... not everyone is fluent enough in Python 
programming to hack the source.  Like me, for example.

>[1] I must say I really don;t get why there is all this jumping up and down
>over what is essentially meta-information in the mail. ANYTHING can be added
>to the headers by mailhosts along the way ... <shrug> if you don;t want to
>look at them then just ignore them .... whats the big deal?

    As others have pointed out, there are cases where one may wish to 
create an "announcements" list under Mailman-- probably because 
that's what a service provider gives a person, and the source is 
beyond their ability to change.  In such one-way lists, the headers 
are largely irrelevant and so are misleading at best.  It might even 
represent a noticeable bandwidth savings to not have them, if 
messages go out with some frequency to a large list.  The 
meta-information headers on the Mailman list I maintain amount to 
just over 0.5KB per message.  Multiple that by 1700 list members and 
about 50 messages per day, and those headers alone represent 42,500KB 
of data per day.  That's a relatively big deal if one is paying for 
the traffic.  And I'm running it via pair.com, so source editing is 
not an option for me.
    Now, I think the headers are very useful for my list and wouldn't 
want to get rid of them, but if they weren't useful then I'd want to 
drop them-- in part to save bandwidth, but also to avoid throwing 
data at the users which doesn't apply.  Without source access, I 
can't do that.  This means that I'm a lot less likely to use Mailman 
for one-way "announcement" lists, or for lists where I think most of 
the traffic will be in short messages.  If I'm limited to using 
Mailman or nothing else, then I might not set up such lists at all.
    Personally, I don't understand why it would be such a big deal to 
offer, as both a Web-based and command-line option, the ability to 
include or omit these headers.  That allows the maintainers to decide 
what's best for their particular list.  I'm sure the option could 
default to "headers on" and nobody would much complain.  Maximum 
flexibility would be afforded by letting admins pick which individual 
headers to keep and which to omit, but I suppose that might be a bit 
much.

--
Eric A. Meyer (eric at meyerweb.com) http://www.meyerweb.com/eric/
Author, "Cascading Style Sheets: The Definitive Guide" and
"CSS 2.0 Programmer's Reference"  http://www.meyerweb.com/eric/books/