You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ant.apache.org by Ja...@rzf.fin-nrw.de on 2003/01/21 15:12:15 UTC
DynamicConfigurator-Interface
When should I use that interface? I had a look at it (and the only
implementing class
I found was XSLTProcess.Factory.Attribute). I don´t see any requirement
for the Attribute-Class to implement that interface. So I ask myself - why
to implement it?
Jan Matèrne
-----Ursprüngliche Nachricht-----
Von: Erik Hatcher [mailto:jakarta-ant@ehatchersolutions.com]
Gesendet am: Dienstag, 21. Januar 2003 14:47
An: Ant Developers List
Betreff: Re: proposed patch to ANT SQL task
Paul,
Also, have a look at the DynamicConfigurator interface. This would
allow you to be completely dynamic about which attributes you accept.
As stated though, be careful about which JDK you are relying on and
that your code is backwards compatible to 1.1.
Erik
Re: DynamicConfigurator-Interface
Posted by Erik Hatcher <ja...@ehatchersolutions.com>.
Unfortunately it was added too late before the 1.5 feature cutoff to
have any implementation within Ant's own codebase. But XDoclet uses
it, which allows for an extensible task facade for XDoclet's subtasks.
A vendor could now write their own XDoclet module and simply dropping
the JAR within the <taskdef> classpath the subtask becomes available...
e.g.
<ejbdoclet...>
<someNewVendorSubTask..../>
</ejbdoclet>
The <ejbdoclet> task implements DynamicConfigurator. Before this
addition, <ejbdoclet> would have to be hardcoded with a
createSomeNewVendorSubTask (or add/addConfigured) method making it very
rigid.
Ant's <ejbjar> task is a perfect candidate for this refactoring also.
The tough part is coming up with the discovery mechanism. The XDoclet
team has a slick way of doing this by looking for XML descriptor files
in the classpath, and plugin modules simply have to embed that within
their JAR.
The same dynamic nature can be done with attributes, so setters for
every attribute are not needed.
<someTask attribute1="..." attribute2="...." ..../>
With no fixed setters for those attributes. Perfect for the extended
JDBC attributes, it seems, although a nested <attribute> element could
be made to do a similar dynamic feature too. It seems cleaner to
express things with attributes with meaningful names though.
Erik
On Tuesday, January 21, 2003, at 09:12 AM, Jan.Materne@rzf.fin-nrw.de
wrote:
> When should I use that interface? I had a look at it (and the only
> implementing class
> I found was XSLTProcess.Factory.Attribute). I don´t see any requirement
> for the Attribute-Class to implement that interface. So I ask myself -
> why
> to implement it?
>
>
> Jan Matèrne
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>