From: Mark Walters <markwalters1009@gmail.com>
To: Peter Wang <novalazy@gmail.com>
Subject: Re: [PATCH 01/18] cli: add stub for insert command
In-Reply-To: <20121119233435.GD2063@hili.localdomain>
References: <1343223767-9812-1-git-send-email-novalazy@gmail.com>
	<87fw46q0f7.fsf@qmul.ac.uk>
	<20121119233435.GD2063@hili.localdomain>
User-Agent: Notmuch/0.14+81~g9730584 (http://notmuchmail.org) Emacs/23.4.1
	(x86_64-pc-linux-gnu)
Date: Mon, 19 Nov 2012 13:52:02 +0000
Message-ID: <871ufppunx.fsf@qmul.ac.uk>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: notmuch@notmuchmail.org
Precedence: list


Hi
On Mon, 19 Nov 2012, Peter Wang <novalazy@gmail.com> wrote:
> On Sun, 18 Nov 2012 17:35:23 +0000, Mark Walters <markwalters1009@gmail.com> wrote:
>> 
>> Hi
>> 
>> I have now been through essentially the whole series (except the tests)
>> and broadly like it. Just to summarise my concerns from the individual
>> replies here:
>
> Thanks for the review.
>
>> 
>> What should insert do when interrupted and is it safe? I am not
>> knowledgeable enough to be confident about that.
>
> Do you mean cleaning up stray files?  Otherwise I think we just return
> an error code, and then it's the caller's responsibility.

It's more that I am concerned that we don't corrupt the database (or get
into some inconsistent state). I have some recollection that there were some
changes to notmuch new to try and make it interrupt safe but I don't
know what they were or whether they apply here.

Best wishes

Mark


>
>> 
>> I think some of the talloc allocations need their return values
>> checked. I am more worried about this here then in most of the rest of
>> notmuch as we are writing to the database (and even to the mailstore
>> itself).
>
> Will do.
>
> Peter