Re: [exim] MUA stats

Edgar Lovecraft wrote:
>
> Messages that are Message-Id based counts do not include any message
> that was found to have an X-Mailer header. Each Message was only
> classified by the first match found.

You should check User-Agent before checking X-Mailer; it's more commonly
used, and many mailers will put identifying details in User-Agent and
cruft in X-Mailer. For example, User-Agent makes it clear that the
messages sent with an X-Mailer of:
> 210 => 'Some Outlooks can't quote properly without this header'
are in fact sent by certain versions/distros of Mutt.

I learned this while starting to do this analysis, but I gave up when my
video driver bluescreened on me. This was the first bluescreen I've
gotten since I built this computer, and decided it was a sign that I'm
not supposed to complete the analysis for some important reason or other.

> This is not going to be 'scientifically' accurate by any means, a
> better solution would be to take into account individual posters..,
> but this should give a good feel of the MUA's used, and Pine wins by
> far (not includeing all of those 'Who_Knows' results ;)

The dedupe by poster is probably a critical step in this analysis; it's
worth noting that Phillip and Tony both use Pine, which probably blows
the Pine numbers way out of proportion.

- Marc