-
1. Re: MC related issue backlog
adrian.brock Jan 13, 2010 5:52 AM (in response to thomas.diesler)Can we create seperate threads for these issues so that proper discussion can occur.
Simply creating a google spreadsheet and assigning ill defined tasks to other people isn't a very good strategy.
Anyway here's my take on the issues (largely guesswork except where I'm already aware of the issue)
JBOSGI-143 - DynamicImport-Package
There are three issues here:
1) The dynamic imports are not getting added to the ClassLoadingMetaData requirements (this is because of number 2)
2) The dynamic imports in jboss-cl don't support wildcards
3) Dynamic imports require some kind of lazy resolve handling when the bundle doesn't automatically progess through to ACTIVE.
(1) is trivial except for the handling of (2) which requires a change in jboss-cl. The PackageRequirement only accepts are a single explict package
as it stands.
(3) requires a change to the deployers to implement the Lifecycle api I defined in JBCL-131. JBCL-131 has only remained open becauseI'm not sure its got all the details correct until the deployers implement it and we can test edge cases.
JBOSGI-204 - Error semantics for OSGi Bundles
I don't know why JBDEPLOY-225 exists? This is an OSGi specific feature. It should implement its own ErrorHandler
to do what it wants. We don't want this semantic for normal deployments.
Why this is assigned to me I don't know? My only input on this topic was pointing out that the ErrorHandler exists. I know very little
about how it works since Ales implemented it. I could figure it out, but then so can you.
JBOSGI-206 - Wire to uninstalled deployments
This has two aspects and the first is actually JBOSGI-213 - Unexpected dependee state changes.
For me this is a bad semantic since it is likely to lead to memory leaks, but that's the OSGi spec so it should be configerd to OSGi bundles.
1) To avoid a bundle getting unresolved when a wire gets unresolved/undeployed then the OSGi requirements need to be able totell the RequirementDependencyItem not to update the dependsOnMe of the dependency. It is that which causes the cascade
of unresolution.
2) When an OSGi bundle gets unresolved. The classloader shouldn't be uninstalled from the classloading system. That should wait for
the refreshPackages(). That means whatever implements the package admin needs to keep track of unresolved classloaders and
properly uninstall them at that time.
The uninstall of the classloader is in the classloading deployers (with an additional safety check in DeployersImpl).
Besides the potential for memory leaks, there's other issues with (2). But any search on classloading FAQs for OSGi would likely
be more comprehensive than any list I can reproduce. ;-)
JBOSGI-209 - Resolver preferences
These are really aimed at resolving rules for ambiguous cases. I don't have a problem with tightening up these rules in jboss-cl
but two of them appear to be about split packages rather than ambiguities.
* A resolved exporter must be preferred over an unresolved exporter.
There is no notion of this in jboss-cl. There will be some preference for previously resolved dependencies
since it resolves first against the ClassLoaderSpace and only then looks at what is in the Domain.
But this is not the exact same semantic since the ClassLoaderSpace can contain unresolved stuff.
However, to me this is a split package and it should resolve to all them with the same version, unless some
rule (e.g. your split package policy) says otherwise.
* An exporter with a higher version is preferred over an exporter with a lower version.This should already work. If it does not then show an example on a bug report.
* An exporter with a lower bundle ID is preferred over a bundle with a higher ID.There is no direct notion of this in jboss-cl, but it should behave in a similar way since the Modules are stored in registration order in a List
inside the Domain. But again this is split package stuff.
NOTE: It is not a split package if they have the same package name but a different version. These are different packages.
-
2. Re: MC related issue backlog
thomas.diesler Jan 13, 2010 8:15 AM (in response to adrian.brock)> Can we create seperate threads for these issues so that proper discussion can occur.
Sure
The strategy that I'd like to use is document in Release Cycle. In case an issue is assigned to you but is actually not on your plate or you have not agreed to make the next progress step, please unassign or reschedule appropriately. An issue may simply be assigned to you because Jira indicates that you are the Project Lead, in which case also please reassign appropriately. Generally, I usually do not assign issues to other people if not asked to do so. Instead, I prefer to make folks aware of an issue and ask them to assign the issue to themselves to make the next progress step.
> JBOSGI-143 - DynamicImport-Package
Follow up on separate thread Initial support for DynamicImport-Package
Assigned to myself to crreate more test coverage
> JBOSGI-204 - Error semantics for OSGi Bundles
This has been resolved
> JBOSGI-206 - Wire to uninstalled deployments
Ok, lets make the next progress in that direction.
> JBOSGI-209 - Resolver preferences
Follow up on separate thread Plugable OSGi Resolver
-
3. Re: MC related issue backlog
thomas.diesler Jan 13, 2010 8:22 AM (in response to thomas.diesler)Adrian, could you please go over the issues that are currently assigned to you and varify that these are actually on your plate and can get done until they are scheduled for? If not, please reassign/reschedule appropriately.