Am Mo., 10. März 2025 um 08:03 Uhr schrieb Martin Monperrus
<martin.monperrus@gnieh.org>:
>
> Hi Carl,
>
> What notmuch could do to support a feature like that is to index all
> quoted content with a different term prefix than it does unquoted
> content. Then, by default search could be made to match on both
> terms.
>
> That would be a great solution, which completely solves the problem.
>

Note that there are some issues with quotes:

First of all, while there are conventions for specifying quotes (such
as "> ") there is no standard, and all libraries which "detect" quotes
can only guess. In particular, this is a problem for messages coming
from the HTML/Outlook/mobile client sector. (As the original author of
"QuoteCollapse for Thunderbird", I know this problem ...)

Second, people use quotes differently. In the above, you quote inline
and selectively (which is what I would do, too). As such, what you
quote is an essential part of your reply and should be indexed as
such. Other people quote everything no matter what, be it because of
corporate policy ("keeping a log") or for lesser reasons. Those are
the quotes which are not "proper content of the reply" and
should/could not be indexed. No library will know the difference.
[Also, GMail fails to insert a 2nd quote level as I see.]

Third, there are those replies (or forwards) where people quote text
to inform you about content which was not addressed to you originally.
Again, you wouldn't want to miss out on that when searching.

As such, I think a solution where you make a deliberate choice at time
of *search* is necessary, and Carl's solution would provide that.

Anecdotally, I suffer from the same type of "quote-heavy" senders as
you seem to - those  who "reply" just to copy addresses, resulting in
monster threads and matches.

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