From: Mark Walters <markwalters1009@gmail.com>
To: Sebastian Schwarz <seschwar@gmail.com>,
	notmuch mailing list <notmuch@notmuchmail.org>
Subject: Re: proposal to remove contrib/notmuch-vim and
 contrib/notmuch-deliver
In-Reply-To:
 <874na0y6x3.fsf@pineapple.i-did-not-set--mail-host-address--so-tickle-me>
References: <87ioygrkhj.fsf@zancas.localnet>
	<uaxa9jsncfm.fsf@beesknees.cern.ch>
	<87a9jsriii.fsf@zancas.localnet>
	<874na0y6x3.fsf@pineapple.i-did-not-set--mail-host-address--so-tickle-me>
User-Agent: Notmuch/0.15.2+269~g01f5508 (http://notmuchmail.org) Emacs/23.4.1
	(x86_64-pc-linux-gnu)
Date: Thu, 05 Sep 2013 10:12:36 +0100
Message-ID: <87r4d3r5rv.fsf@qmul.ac.uk>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Precedence: list


On Wed, 04 Sep 2013, Sebastian Schwarz <seschwar@gmail.com> wrote:
> Hello,
>
> On 2013-04-09, David Bremner <david@tethera.net> wrote:
>> David Belohrad <david@belohrad.ch> writes:
>>
>>> is actually someone using notmuch deliver?
>
> I am using notmuch-deliver with maildrop.  This allows me to
> filter and tag based on any header field of the message.
>
>>> [...]
>>
>> notmuch-deliver is replaced by notmuch insert
>
> Because of the way notmuch-insert(1) deals with failure it is
> entirely useless to me.  From the manual page:
>
>> This command returns exit status 0 if the message was
>> successfully added to the mail directory, even if the message
>> could not be indexed and added to the notmuch database.
>
> This means that if the index is locked when a message is
> delivered it will not be tagged appropriately.  The message will
> still be written to disk.  However if you rely on the tags to
> find the message it effectively gets lost.
>
> notmuch-insert(1) should fail if it isn't able to successfully
> perform all of the following three tasks:
>
> - write the message to disk
> - index the message
> - tag the message
>
> It should return with a non-0 exit status and let its caller --
> the MDA or another notmuch client -- handle the failure, since
> notmuch-insert(1) cannot make any guarantees by itself anyway.
> This would usually make an MDA retry delivering the message at a
> later time.

I think this is a good reason not to remove notmuch-deliver until there
is at least an option to insert to fail in this case. A step in this
direction was made in
id:1374365254-13227-1-git-send-email-novalazy@gmail.com but this is not
a complete solution and is not in mainline anyway.

Best wishes

Mark


>
> Regards,
> Sebastian
> _______________________________________________
> notmuch mailing list
> notmuch@notmuchmail.org
> http://notmuchmail.org/mailman/listinfo/notmuch