Re: [tycho-user] Tycho packaging: Maven groupId and artifactId vs OSGI B

[Ben Zegveld <benzegveld@xxxxxxxxx>]

Hey,

* What's your issue with using artifactId == Bundle-SymbolicName in practice?

2 reasons that come to mind right now:
1. Our maven build OSGI bundles all use groupId.artifactId as Bundle-SymbolicName, but this we could change if needed.

2. If we would change this to just the artifactId then the product would contain (among others) the following plugins:

domain (6.11.0)

domain (7.1.1)

domain (1.0.2)
all from different components that come together in a single product. Well it makes it kinda hard to see if I have the correct bundles.

I do debug from time to time from eclipse, and our product is an eclipse RCP application.

* Did you try to simply skip the execution execution of the validation mojo? If so, how far does it go without complaining?
It would probably build without problems but a side-effect would be that the tycho-versions-plugin would not update the manifests.

And when making a release this plugin is used in the jenkins job we have.

Ben Zegveld

On Wed, Sep 28, 2016 at 2:48 PM, Mickael Istria <mistria@xxxxxxxxxx> wrote:

Hi,

On 09/28/2016 02:18 PM, Ben Zegveld wrote:

That got me wondering why tycho expects ArtifactId:Version while this is not by definition unique.

I believe it's caused by p2 (the dependency resolve for OSGi bundles in Eclipse) doesn't have a concept of groupId at all. Also, the typical OSGi bundle using a fully-qualified name such as `org.eclipse.core.runtime_3.12.0.v20160606-1342.jar` when Maven usually uses `${artifactId}-${version}.jar`. So it seemed more straightforward and natural to map Bundle-SymbolicName on artifactId.

Is there a way to configure Tycho to expect the combination of groupId and artifactId as the Bundle-SymbolicName?

Not that I am aware of, but you could consider creating an enhancement request to support that practice as well.

However, some preliminary questions:
* What's your issue with using artifactId == Bundle-SymbolicName in practice?
* Did you try to simply skip the execution execution of the validation mojo? If so, how far does it go without complaining?

HTH

--
Mickael Istria
Eclipse developer for Red Hat Developers
My blog - My Tweets

_______________________________________________
tycho-user mailing list
tycho-user@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/tycho-user