Cheers,
Steven Harris
Director of Engineering
[EMAIL PROTECTED]
www.terracotta.org
On May 22, 2007, at 3:04 PM, Geert Bevin wrote:
Good question. If someone were to serialize that graph and de-
serialize it.
Is that case handled?
Afaik, this isn't automatically handled through serialization.
If it isn't, can we at least cover a wider subset of the world
without naming classloaders?
With web apps for instance? If not, what we would need to make
things simpler.
I sort of fear that this might be something that is different on a
case-by-case basis. Very dependent on the applications and frameworks.
I'll bet we can come up with some default mode that handles a nice
swath of the world and then rely on named classloaders for the rest.
but that is just a theory at this point.
Would be nice to be able to alias classloader names in config to
make it easier to share
objects between say tomcat and a swing app?
I agree that this could be nice, but wouldn't it make the whole
thing even more complex, ie. another concept to understand:
"aliased named classloaders"?
For the people who need this the only other option is to do it
manually which is a big pain. For everyone else, they would never
know the concept exists.
Just being able to make classloader names equivalent between nodes
would remove a whole bunch of ugly code in cross appserver and cross
classloader
architecture worlds. We should probably pick some use cases. i.e.
osgi, tomcat to swing, etc. Look at them one at a time and figure out
how to best balance
out form over function.
Cheers,
Steven Harris
Director of Engineering
[EMAIL PROTECTED]
www.terracotta.org
On May 22, 2007, at 2:52 PM, Geert Bevin wrote:
1. object X instantiates an object Y by loading the class
explicitly through another classloader
2. object Y is stored into object Z
3. object Z is shared
How can object Z on another node know which classloader has to be
used for object Y's class?
On 22 May 2007, at 23:48, Steven Harris wrote:
I believe so. Question in my mind is, why can't we do the same.
Cheers,
Steven Harris
Director of Engineering
[EMAIL PROTECTED]
www.terracotta.org
On May 22, 2007, at 2:46 PM, Geert Bevin wrote:
Doesn't serialization just use the caller's current classloader?
On 22 May 2007, at 23:42, Steven Harris wrote:
I think we can follow the same rules that serialization follow
about choosing classloaders but I'm not sure.
Cheers,
Steven Harris
Director of Engineering
[EMAIL PROTECTED]
www.terracotta.org
On May 22, 2007, at 2:38 PM, Geert Bevin wrote:
What about classes that are loaded through another classloader
(ie. an application having several classloaders). How well TC be
able to automatically tell which classloader it has to
retrieve it
from, if it isn't named?
On 22 May 2007, at 23:22, Steven Harris wrote:
This is something that has come up a few times over the last
few
years. Alex mentioned a solution and gkeim mentioned a similar
solution.
I want to schedule a brain storming session on it in the coming
weeks
but...
Since I don't believe we ever push objects to a client
anymore, at
least not all the way to instantiation, I believe we can
just use
the
classloader of requesting client. This may not
have always been true but I think it is true now. What reasons
can we
think of that would require us to need named classloaders
anymore?
--
Geert Bevin
Terracotta - http://www.terracotta.org
Uwyn "Use what you need" - http://uwyn.com
RIFE Java application framework - http://rifers.org
Music and words - http://gbevin.com
_______________________________________________
tc-dev mailing list
[email protected]
http://lists.terracotta.org/mailman/listinfo/tc-dev
