Paul Wise <pabs3@bonedaddy.net> writes:


> that changes the meaning of the query but I couldn't find any way to
> prevent an arbitrary Message-Id from inserting parentheses into the
> query. Also notmuch converts Message-Id (test)@hostname to just
> @hostname in the Xapian database. I think that right now I am not sure
> that re-implementing all the idiosyncrasies of notmuch queries and
> Message-Id munging within notmuch-mutt is the way to go, especially
> since Message-Id fields that intersect with notmuch features are rare.

The code that does removal of spaces and () delited sequences goes back
to Carl's 2009 implementation of manual header parsing (which we have
mostly, but not entirely replaced with calls to gmime).

It would probably be reasonable to disable that code (or maybe migrate
to gmime parsing of message-ids?), but the cost/benefit analysis is not
too clear to me.

I had to look this up, but apparently parens mark a comment in RFC822 /
RFC5322. If I understand RFC5322 correctly then comments are only
permitted in the obsolete syntax, but I guess it makes sense to accept
those. I don't know how many messages in the wild use this syntax;
of the .01% in in my mail store, most don't put the comment inside the actual <>
delimited message-id id, and those that do look like errors of various kinds
- some unbalanced parens
- some (null) indicating printf-ing null pointers
_______________________________________________
notmuch mailing list -- notmuch@notmuchmail.org
To unsubscribe send an email to notmuch-leave@notmuchmail.org