Robert Burrell Donkin ha scritto:
On Mon, May 12, 2008 at 3:03 PM, Stefano Bagnara <[EMAIL PROTECTED]> wrote:
There are probably many "levels" of OSGi support:
1) Create a single bundle for the whole JAMES Server
2) Create one bundle for each function
3) Create one bundle for each library
4) Create one bundle for each module
AIUI libraries used by OSGi require appropriate MANIFEST entries on a
per jar basis
Didn't know this. In this case we are forced to fix it for each module.
IMHO the really needed things would be to be able to undeploy the whole
spoolmanager (and by result all the processors/mailets) and redeploy an
updated version. In the mean times JAMES would be fully working because the
incoming channel do not block (because of the spool existence). So I think
the "level" that will bring us more advantages with less work in #2. Of
course #4 would be good, but it won't add too much to #2. #1 alone does not
give many advantages over what we already do with current standalone james.
I don't really know how complex would be to create a single bundle for each
of our functions and then to have an OSGi-deployment that do something
similar to the spring-deployment method (by emulating an avalon container
for OSGi).
I read something about OSGi and if IIRC OSGi would be very unhappy when the
same package exists in 2 different bundles, and we know we have a bit of
chaos about packages. Is this a blocker?
OSGi does indeed forwn upon this
i wasn't thinking of using OSGi to assemble the whole of JAMES ATM,
just mailets. configuration is the main issues with mailets but i
wonder whether the specification should really concern itself with
configuration at all. i think JAMES would benefit from supporting new
mailet loading engines. OSGi enabling our mailets would open up the
possibility of using OSGi as a mailet loader.
I don't understand this.
Can you explain how to use OSGi to load mailets?
(e.g: How can you reload/reconfigure mailets if you don't unload the
spoolmanager?)
Stefano
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
