Thread context class loader...

[Genevski, Pavel]

Hi,

Do you have any comments on my suggestions below ?
And in addition, can you point me to the source code locations of the
reference implementation of this JSR ? 


Best Regards,
Pavel Genevski
Deployment Team
SAP Labs Bulgaria
Tel. +359 2 9157 133

-----Original Message-----
From: Java Community Process JSR #277 Expert List
[mailto:JSR-277-EG at JCP.ORG] On Behalf Of Genevski, Pavel
Sent: Wednesday, 17. October 2007 11:41
To: JSR-277-EG at JCP.ORG
Subject: Re: Thread context class loader...

Hi,

While we are talking about classloading I would like to add a few
comments about the cyclic dependencies mentioned in section 8.3.4 of the
spec. How do you imagine the classloading with cyclic dependencies? What
are the benefits of allowing cyclic dependencies ? 

In my opinion cylcic dependencies occur mostly because of bad system
architecture and are always hard to handle. The tough questions that
emerge with them are for example:

- what should be built first
- which classloader should be created first. 

One obvious runtime solution for classloading of modules with cyclic
dependencies is to have a common classloader for all the modules in the
cycle, but are there enough benefits that come from cyclic dependencies
that are worth doing this and breaking module isolation?


Best Regards,
Pavel Genevski
Deployment Team
SAP Labs Bulgaria
Tel. +359 2 9157 133



-----Original Message-----
From: Java Community Process JSR #277 Expert List
[mailto:JSR-277-EG at JCP.ORG] On Behalf Of Andy Piper
Sent: Tuesday, 18. September 2007 12:44
To: JSR-277-EG at JCP.ORG
Subject: [LIKELY JUNK]Re: Thread context class loader...

At 03:55 18/09/2007, Bryan Atsatt wrote:
>Thoughts?

Given our experience with OSGi I agree that this needs addressing. We
have found, though, that the notion of a "thread of execution" is
somewhat more amorphous in a non-JEE environment. In using
Spring-OSGI we have found we often have to force the client
classloader to be a delegate of the target bundle's, but there are
issues with trying to actually make this switch happen. I'm wondering
whether we actually need some VM-level support to make this work well.

andy


Notice:  This email message, together with any attachments, may contain
information  of  BEA Systems,  Inc.,  its subsidiaries  and  affiliated
entities,  that may be confidential,  proprietary,  copyrighted  and/or
legally privileged, and is intended solely for the use of the individual
or entity named in this message. If you are not the intended recipient,
and have received this message in error, please immediately return this
by email and then delete it.