One of the annoyances with notmuch-address (and motivations for
various third party work-a-likes) is the fact that it does not output
data in (descending) frequency order.  This series adds that for
either --output=count or --deduplicate=address. This is because the
current --dedup=mailbox handling does not actually count the
occurences.

I see two ways to go here

1) Unify the handling of --dedup={mailbox,address}. This will probably
yield a performance hit for dedup=mailbox. How big probably depends on how many duplicates you have.
The main difference would be that the output would all arrive at the end, like with --dedup=address.

If people are using --dedup=mailbox for large queries, now is the time
to speak up. I noticed the builtin completion in notmuch-emacs is
already using --deduplicate=address, so probably no change there.

2) Apply something like this series, and deal with questions from
users confused by why only one of the deduplication options, or
output=count, sorts.

At least in my tests, there does not seem to be a noticable
performance hit for the current series. I'm also interested to know if
other people do notice such a hit.


_______________________________________________
notmuch mailing list -- notmuch@notmuchmail.org
To unsubscribe send an email to notmuch-leave@notmuchmail.org