From: "Peter Donald" <[EMAIL PROTECTED]>

> On Sat, 23 Feb 2002 21:33, Jose Alberto Fernandez wrote:
> > From: "Peter Donald" <[EMAIL PROTECTED]>
> >
> > > If the set of classes from each source is disjoint then there is no real
> > > need for a strictly hierarchial approach. You can easily accomodate the
> > > multiparent scenario.
> > >
> > > However what you describe does not need this and may be modeled as simple
> > > aggregation of classloaders like most other container based APIs specify.
> >
> > So what is your point?
> 
> That nothing needs to be changed as all is good now.
> 

Are you trying to say that the <style> <ejbjar> etc. task are working as 
advertised, 
and that it should be just natural that people need to pick appart the jars 
that we shipped
to get their work done?

> Myrmidon just declares dependencies via the JDK1.3 
> "Optional Package" spec on libraries and then all dependencies will loaded 
> into same classloader (unless they already exist in parent ClassLoader).
> 

This is also available in Antlib proposal. However, in my experience ans due to 
the fact that many 3rd party libraries have a different release cycle than ANT,
you finish with libraries with the wrong name, and it is difficult to use 
libraries
relative to the project.

> > Unless you are planning to write some new proxy classloaders that allow for
> > multiple parents, I really do not understand yout comments above.
> 
> They are already written but I don't think they are appropriate for ant. I 
> think you will find that reusing classloaders rather than recreating a new 
> one will cause far more troubles than its worth and not at all viable in some 
> cases (ie Velocity case).

Can you please elaborate? What is the Velocity case?

Jose Alberto



--
To unsubscribe, e-mail:   <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>

Reply via email to