MIME-Version: 1.0
Sender: polatel@gmail.com
In-Reply-To:
 <CAMP44s3SyU4WVV0_McHWseNL=jmMnAXO2EdZK4Xk-wrCHPVD8A@mail.gmail.com>
References: <1335185032-13075-1-git-send-email-felipe.contreras@gmail.com>
	<CADv3eywAvyMuh3vWLwyuf0Ui_kskwp9875pGxCR1GTm7deN9Pg@mail.gmail.com>
	<CAMP44s3SyU4WVV0_McHWseNL=jmMnAXO2EdZK4Xk-wrCHPVD8A@mail.gmail.com>
Date: Mon, 23 Apr 2012 23:43:04 +0300
Message-ID:
 <CADv3eywb0tguYowTAK5Ag9YZ48zFZA0QJVNEj_cZcCpr-76Bbg@mail.gmail.com>
Subject: Re: [PATCH] ruby: make sure the database is closed
From: Ali Polatel <alip@exherbo.org>
To: Felipe Contreras <felipe.contreras@gmail.com>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Cc: notmuch@notmuchmail.org
Precedence: list

2012/4/23 Felipe Contreras <felipe.contreras@gmail.com>:
> On Mon, Apr 23, 2012 at 5:04 PM, Ali Polatel <alip@exherbo.org> wrote:
>
>> I'd rather not do this.
>> Please read: http://comments.gmane.org/gmane.comp.lang.ruby.general/3203=
24
>
> OK, I've read this.. So?

You are one step close to what I thought you had thought.

> The order in which Ruby's garbage-collector frees the database and
> other objects is irrelevant, because with this patch we are not
> manually freeing other objects, only the database.

What I wanted was to make all objects "depending" on the database to be unu=
sable
once the database object is freed. Seeing that's not practically easy
I decided to leave
it to the user.

> Sure, it's _better_ if the user calls close(), even better if it's
> inside an 'ensure', and even better if blocks are used (which I am
> using in most cases), but that's not *required*.

If you have such a use case, I'm fine with that patch.
I might push it in the next few days or get someone else to push it.

> The user might just do:
>
> def foo
> =A0db =3D Notmuch::Database.new($db_name, :mode =3D> Notmuch::MODE_READ_W=
RITE)
> end
>
> That's perfectly fine in Ruby (although not ideal), since 'db' will
> get garbage-collected. But nobody will be able to use the database
> again until that process is killed.
>
> You think that's correct?

Yes that is correct. I have not thought about this.
I'd say it's a partial misunderstanding on my part due to lack of
(and/or too much) vodka.

> --
> Felipe Contreras

        -alip