-----Ursprüngliche Nachricht-----
Von: marc fleury [mailto:[EMAIL PROTECTED]]
Gesendet: Mittwoch, 20. Juni 2001 05:44
An: Jboss-Development@Lists. Sourceforge. Net
Betreff: [JBoss-dev] micro-kernel design



>Dr JUNG, here is one for you... I realize that right now the base is very
>static in the sense that we cannot really extend the MLet beyond the
>initialization (ClassPathExtension relies on the contextCL).  What that
>means is that if you really want to change the implementation of something
>in the system at run time well... you can't.

>The main weakness is the reliance on the context class loader to get the
>MLet, it assumes that we come with the system thread.  I.e. there is no way
>to come from the html adaptor and say "deploy this datasource from this
>http://myserver/datasource.xml"; and that is really static... very far from
>what I want for the 3.0 vision.

>long story short, we were exploring the class loader stuff for the
>applications only but it turns out that the first application that is
>dynamic and needs to have parts of it cycled is the JBOSS-SYSTEM *itself*
:)
>so i will try to tackle this soon.  Without this we really don't have a
real
>micro-kernel design with 24x7 uptime.

That´s what I tried to introduce with the "abstract" deployer code that I
sent 
around last Christmas (laast Christmas, I gave you my heart, 
but the very next day, ...) and that nobody really did understand.

In fact, if we could lift the logic that is now located in the
J2eeGlobalDeployer (or say, in the J2eeGlobalDeployer before Mad Andy did
that mad monster change which simply deleted all the lines ;-) underneath
the J2eeDeployer at a level above, there is a chance
that we could also derive some org.jboss.jmx.JmxDeployer, don´t you think
so?

Best,
CGJ


_______________________________________________
Jboss-development mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/jboss-development

Reply via email to