v5 of regexp searching

David Bremner david at tethera.net
Thu Feb 16 19:07:48 PST 2017


The first 3 patches are general cleanup of the way that the query parser is configured.

[patch v5 1/6] lib: merge internal prefix tables
[patch v5 2/6] lib: Let Xapian manage the memory for FieldProcessors
[patch v5 3/6] lib: create field processors from prefix table

This is id:20170121135917.22062-1-david at tethera.net, revised with
Tomi's comments, and error handling for missing trailing
delimiter. For the moment missing initial delimiter is ignored for
future compatibility with paths.

[patch v5 4/6] lib: regexp matching in 'subject' and 'from'

The next two are RFC. We discussed several options for dealing with
the fact that / is a valid message-id character. Here the idea is to
used a separate prefix mid: for potential regexp searches. Jani had
previously suggested trying both the regexp and non-regexp form of the
query. That might be a more DWIM ui, but it would have the potentially
surprising side effect of having id: match more than one message.

The first four patches can be merge earlier, if we want to continue
thinking about this UI (or just drop the last two patches).  Also, the
id case needs doc and tests.

[patch v5 5/6] lib: add mid: as a synonym for id:
[patch v5 6/6] lib: Add regexp searching for mid: prefix


More information about the notmuch mailing list