Mailing-List: contact dev-help@excalibur.apache.org; run by ezmlm
Precedence: bulk
Reply-To: "Excalibur Developers List" <dev@excalibur.apache.org>
Received-SPF: neutral (hermes.apache.org: local policy)
Message-ID: <33336.134.91.35.75.1111772771.squirrel@webmail.paiwastoon.com.af>
Date: Fri, 25 Mar 2005 18:46:11 +0100 (CET)
Subject: Rountrip for configurations?
From: "Torsten Schlabach" <tschlabach@apache.org>
To: dev@excalibur.apache.org
Cc: jwkaltz@apache.org
Reply-To: tschlabach@apache.org
User-Agent: SquirrelMail/1.4.2
MIME-Version: 1.0
Content-Type: text/plain;charset=iso-8859-1
Content-Transfer-Encoding: 8bit
Importance: Normal

Dear all,

I have a quite simple question:

Is a component configuration (some .xconf) meant to be a one-way street?
Is the basic idea that a component is configured from some .xconf file
that got written manually by a human beeing?

We would have a need for roundtrip configurations in the sense that the
component comes up, get's its initial configuration from the .xconf file,
then maybe later changes configured values from the Java code and wants to
save it back to the .xconf.

The only way I have seen this implemented is by having a save() method in
the component which will built the entire configuration from scratch an
serialize it. I think this is not only a cumbersome but also dangerous
approach as it can easily happen that some options in the .xconf file just
get lost because they might not be read and written back.

I wonder if it would be a good idea to move to a JDOM like approach which
would be to instantiate the .xconf in memory, change spefific elemnts and
write it back.

Is there already something like that?
Am I on the entirely wrong path?

Regards,
Torsten


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@excalibur.apache.org
For additional commands, e-mail: dev-help@excalibur.apache.org