[jboss-dev-forums] [Design of OSGi Integration] - Re: JBOSGI-38 and bundle deployment order

[alesj]

I still don't see what this has got to do with RBDI?

I wouldn't worry about such cases:
* deploy A which depends on X
* wait 30min
* deploy X
Since this is not realistic.
And even then, I don't see a problem, 
as the only thing we'll get is a exception meant to be a warning.
You either deploy the missing pieces or you remove the initial deployment.

As discussed, mostly it's this behavior that we see/use:
* MD::addDeployment A
* MD::addDeployment X
* MD::process
* MD::checkComplate

And even with this and current HD scanner we might 
not get all dependent deployments in one group/snapshot,
e.g. X is just being droped, while we did a scan,
but that wouldn't be a problem apart from IDE,
which can easily be dealt with - ignored or handled properly.

If we're gonna use MC's VDF to handle our OSGi Runtime deployments,
we should do it as we're already doing them - with proper dependencies tied into deployment's lifecycle.
And that's exactly what RBDI and a proper deployer's split does.

View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4215149#4215149

Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4215149