Message-ID: <4ba237eb.0d67f10a.0fba.0105@mx.google.com>
Date: Thu, 18 Mar 2010 15:27:12 +0100
From: Sandra Snan <sandra.snan@handgranat.org>
To: notmuch <notmuch@notmuchmail.org>
User-Agent: Wanderlust/2.14.0 (Africa) SEMI/1.14.6 (Maruoka) FLIM/1.14.9
	(=?UTF-8?B?R29qxY0=?=) APEL/10.7 Emacs/23.1 (x86_64-pc-linux-gnu)
	MULE/6.0 (HANACHIRUSATO)
References: <4ba01c54.9298cc0a.66cd.7695@mx.google.com>
In-Reply-To: <4ba01c54.9298cc0a.66cd.7695@mx.google.com>
MIME-Version: 1.0 (generated by SEMI 1.14.6 - "Maruoka")
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
Subject: Re: [notmuch] Tag search peculiarities
Precedence: list

Hey, Ben.
Ben Gamari <bgamari.foss@gmail.com> wrote:
> notmuch tag -new tag:new and tag:list notmuch tag -new +inbox tag:new

Is there a new line between the calls? Like:
notmuch tag -new tag:new and tag:list # removes new from list
notmuch tag -new +inbox tag:new # replaces those that still are new with in=
box

> However, I found that mailing list traffic was still getting through.

What do you mean by =E2=80=9Cthrough=E2=80=9D? Do you mean that there still=
 are
messages tagged both inbox and list? Even though you can=E2=80=99t search f=
or
them?

If a message is tagged both inbox and list, but not new then the two
lines of your script that you posted so far wouldn=E2=80=99t change it, and=
 it
would still bo both inbox and list.

It=E2=80=99s risky being so dependent on the new tag.

> After investigating further, I found that any query in the form of "tag:i=
nbox
> and tag:$TAG" would return no results. Strangely, all other combinations =
of tag
> searches (i.e. "tag:lkml and tag:unread") seem to work just fine.
>=20
> Has anyone else noticed this sort of behavior? Does the inbox tag have so=
me
> special signifigance that I should know of?

Yes, many versions of the reading client (for example, the version of
notmuch.el that ships with the notmuch package in debian) remove the
inbox tag from messages once you scroll past them or when you press
the key that=E2=80=99s bound to notmuch-show-archive-thread (default is =E2=
=80=9Ca=E2=80=9D).

> Is my index just FUBAR? (the ladder would be very strange
> considering it's only a few days old and I can't think of any
> crashes, etc. that might have corrupted it) Any ideas for debugging?

Add the tags manually to a few messages, search for them again, and be
sure to not run your tagging script while looking at this problem.

Michal=E2=80=99s tip, looking at a few lines of notmuch dump, is a good idea
too. Notmuch dump is fast, so don=E2=80=99t be afraid (notmuch restore on t=
he
other hand=E2=80=A6 but even that completes fast enough.)

I=E2=80=99ve often been surprised at weird tag situations then realized that
it was old versions of my own scripts that had ran in the background
and I=E2=80=99ve forgotten about it.

Boolean algebra and set theory is tricky stuff! Good luck.

Sandra