quirks with subject searching

quirks with subject searching

[Matthew Lear]

Hello all,
This may well be due to my novice use of search syntax, but I'm having
issues find needle in haystack within the subject of an email thread.
This simple example highlights the problem I'm having:

When using notmuch search subject:foo for threads with the following
subjects...

item: foo
an item: foo
an item: XYfoo1234
an item: XYfoo bar

... only the first two are found.

Also:
* notmuch searching for subject:"item: f" gives zero results.

* notmuch searching for subject:"(*foo*)" results in the first two threads
being found and also a thread with the word footprint in the middle of the
subject being found (subject is "memory footprint without feature").

* notmuch searching for subject:"an item: XYfoo1234" results in the third
being found.

Why doesn't notmuch search subject:foo find all four threads, and why
doesn't notmuch search subject:"item: f" find anything at all?

This was spurred on by me having a thread with the following as the subject..

Release Update: BT52_ETW_MD4.5.4alpha1_take1

..and notmuch search subject:4.5.4 finding nothing.

Something doesn't seem right here. man notmuch-search-terms doesn't seem
to explain this as far as I can see.
FWIW I'm using notmuch from git, master @ 294bb6d.
Any thoughts / tips?
Thanks,
--  Matt

Re: quirks with subject searching

[David Bremner]

Matthew Lear <matt@bubblegen.co.uk> writes:
>
> item: foo
> an item: foo
> an item: XYfoo1234
> an item: XYfoo bar
>
> ... only the first two are found.
>
> Also:
> * notmuch searching for subject:"item: f" gives zero results.

notmuch (and xapian) are based on searching for words, not
substrings. It only finds subwords through a process of stemming [1]. 

>
> * notmuch searching for subject:"(*foo*)" results in the first two threads
> being found and also a thread with the word footprint in the middle of the
> subject being found (subject is "memory footprint without feature").

wildcards [2] are supported only at the end of words, so I suppose the first
* is just ignored.

>
> Why doesn't notmuch search subject:foo find all four threads, and why
> doesn't notmuch search subject:"item: f" find anything at all?

Hopefully the above clears it up. This topic comes up fairly often; I'm
not sure if there is something we could add to the  (already rather
long) notmuch-search-terms manpage that would help.


[1] "Stemming" in notmuch-search-terms (7)
[2] "Wildcards" in notmuch-search-terms (7)