From: Jameson Graef Rollins <jrollins@finestructure.net>
To: Mark Walters <markwalters1009@gmail.com>,
	Notmuch Mail <notmuch@notmuchmail.org>
Subject: Re: [PATCH 5/8] emacs: modify search tag functions to use new
	notmuch-tag interface
In-Reply-To: <871unwn40g.fsf@qmul.ac.uk>
References: <1333354853-25729-1-git-send-email-jrollins@finestructure.net>
	<1333845338-22960-1-git-send-email-jrollins@finestructure.net>
	<1333845338-22960-2-git-send-email-jrollins@finestructure.net>
	<1333845338-22960-3-git-send-email-jrollins@finestructure.net>
	<1333845338-22960-4-git-send-email-jrollins@finestructure.net>
	<1333845338-22960-5-git-send-email-jrollins@finestructure.net>
	<1333845338-22960-6-git-send-email-jrollins@finestructure.net>
	<871unwn40g.fsf@qmul.ac.uk>
User-Agent: Notmuch/0.12+112~gffb05d6 (http://notmuchmail.org) Emacs/23.4.1
	(x86_64-pc-linux-gnu)
Date: Sat, 14 Apr 2012 12:07:54 -0700
Message-ID: <871unq169h.fsf@servo.finestructure.net>
MIME-Version: 1.0
Content-Type: multipart/signed; boundary="=-=-=";
	micalg=pgp-sha256; protocol="application/pgp-signature"
Precedence: list

--=-=-=

On Mon, Apr 09 2012, Mark Walters <markwalters1009@gmail.com> wrote:
> My only comment here is that I wonder about the tag-all command. I can
> see that this is a performance win over applying tag region to the whole
> buffer but is it another place where we have a race with incoming mail
> (ie we tag everything that now matches rather than everything in the
> buffer).

Yes, I think this is a problem.  notmuch-{search,show}-tag-all are both
subject to race conditions.  I suggest we remove notmuch-search-tag-all
in favor of the region tagging capabilities of notmuch-show-tag.  I will
submit a patch for that.  notmuch-show-tag-all is trickier, since there
no tag-by-region capability at the moment.  That will have to be added
before this function can be replaced.

> A second advantage with the tag-region form is that it automatically
> does excludes. Currently notmuch-tag does not do excludes as it looked
> very hairy to implement correctly: thus the tag-all command could spend
> a lot of time tagging excluded messages.

Actually, I don't think this is true, at least for search mode.  In
search mode threads are tagged by their thread id, therefore any
excluded messages in the thread will receive the same tagging
operations.

The current operation is at least, in my opinion, self consistent.  If
we're going to tag by thread then all messages in the thread should be
tagged, whether or not they're excluded.  I do think, however, that it
might be nice to add a tagging function that only tags the matching
messages in the thread.  In that case excluded messages would not be
tagged.  That's for future work, though.

jamie.

--=-=-=
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)
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=q9b+
-----END PGP SIGNATURE-----
--=-=-=--