From: David Bremner <david@tethera.net>
To: Anton Khirnov <anton@khirnov.net>, notmuch@notmuchmail.org
Subject: Re: [PATCH v5 0/6] indexing attachment contents
In-Reply-To: <20260815140127.26199-2-anton@khirnov.net>
References: <20260815140127.26199-2-anton@khirnov.net>
Date: Sat, 15 Aug 2026 14:36:16 -0300
Message-ID: <874igv9tgv.fsf@tethera.net>
MIME-Version: 1.0
Message-ID-Hash: DGBOXJWL6EGJ5O77WGYLWHCEPZJM32BB
Precedence: list
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit

Anton Khirnov <anton@khirnov.net> writes:

> Hi,
> another iteration of the set adding support for indexing attachment
> contents. I believe I've addressed all the review comments from the
> previous round [1].
>
> The first patch now contains a simple performance test that runs reindex
> on all emails with attachments - twice with no filter, and once with a
> trivial /bin/true filter. By placing the test early in the series it is
> easy to see that the following patches do not make the filter-less runs
> slower.

Thanks for humouring me and doing the performance test.

>
> As for the run with the filter, the test does show that the fork
> overhead is susbtantial, especially with larger database sizes. As per
> previous discussion I believe the cost is acceptable, since the
> alternative (a persistent filter daemon) would be far more complex
> with
> regards to both implementation and administration.

Yeah, I think it's a tradeoff individual users can make.

>  doc/examples/README.filter.rst              |  23 ++
>  doc/examples/bwrap.sh                       |   9 +
>  doc/examples/filter.py                      | 385 ++++++++++++++++++'

It seems that doc/examples/firejail.profile went missing at some point
(also missing in v4)

Other than that looks good (I'll need to update the debian packaging,
to install the examples but that's a me problem).
_______________________________________________
notmuch mailing list -- notmuch@notmuchmail.org
To unsubscribe send an email to notmuch-leave@notmuchmail.org