Mailing-List: contact dev-help@couchdb.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@couchdb.apache.org
Received-SPF: neutral (athena.apache.org: local policy)
Content-Type: text/plain; charset=windows-1252
Mime-Version: 1.0 (Apple Message framework v1082)
Subject: Re: scripts in [update_notification] section ?
From: Jan Lehnardt <jan@apache.org>
In-Reply-To: <AANLkTim8VMbErmbkzKnuDS3-jW+DKq5HNhb4Vg7YQgK9@mail.gmail.com>
Date: Mon, 15 Nov 2010 17:12:37 +0100
Content-Transfer-Encoding: quoted-printable
Message-Id: <121768C2-28F2-4241-ADF5-EC8D961C7670@apache.org>
References: <4CE1512D.5010607@free.fr>
 <AANLkTim8VMbErmbkzKnuDS3-jW+DKq5HNhb4Vg7YQgK9@mail.gmail.com>
To: dev@couchdb.apache.org


On 15 Nov 2010, at 16:58, Zachary Zolton wrote:

> I think your update notification script is not supposed to exit every
> time there's a change. It's supposed to stay running in a loop where
> you get the next line from standard input (for each update), only
> quitting when that stream gets closed.

That is correct.

> That said, the update notification feature should generally be
> considered deprecated by the newer changes API:
> http://guide.couchdb.org/draft/notifications.html

That is not correct :) =97 The two APIs serve different purposes and
none of them is deprecated.

That said, a /_db_update handler that acts like DbUpdateNotification
but over HTTP would be neat.

Cheers
Jan
--=20


>=20
> Cheers,
> Zach
>=20
> On Mon, Nov 15, 2010 at 9:26 AM, cdr53x <cdr53x@free.fr> wrote:
>> Hi,
>>=20
>> I read the article concerning the view updates in the wiki
>>  (http://wiki.apache.org/couchdb/Regenerating_views_on_update) and =
tried to
>> test this.
>>=20
>> I just put a basic script  that simply logs a message in my conf.ini =
:
>>=20
>> [update_notification]
>> view_update=3D/home/tests/couchdb/deployed/bin/update_views.rb
>>=20
>> After restarting couch, I noticed that this script is invoked _all_ =
the
>> time, i.e as soons as the script finishes couch starts another one. =
And no
>> insertions/updates are performed on the db.
>>=20
>> So there seems to be no relation between the script execution and the =
fact
>> that a document has been inserted or updated.
>>=20
>> Is this the expected behavior or a bug ?
>>=20
>> Regards,
>>=20
>> cdrx
>>=20
>>=20