|
| Re: Target platform vs. Buckminster headless installation [message #499889 is a reply to message #499813] |
Tue, 24 November 2009 13:37   |
|
Hi Florian,
DI Florian Hackenberger wrote:
> Hi Thomas,
>
> The problem with the target platform turned out to be related to the problem
> I reported here a while ago: "RMAP vs. buckminster headless, who wins the
> component resolution battle?".
>
There should never be a battle. Buckminster consults the target platform. The target platform is the
runtime by default. If it isn't, then the runtime is never consulted.
> The question is: How can I specify that buckminster headless should resolve
> components using the target platform definition I imported, but not the
> buckminster headless installation? Unfortunately buckminster chooses the
> component with the highest version from either the target platform or the
> buckminster installation. In the original thread you suggested to add
> advisory nodes like this:
>
> <cq:advisorNode namePattern=".*" componentType="eclipse.feature"
> useTargetPlatform="false"/>
> <cq:advisorNode namePattern=".*" componentType="osgi.bundle"
> useTargetPlatform="false"/>
>
> Unfortunately, that disables both the target platform, as well as the
> buckminster installation as sources for features and bundles.
This disables the target platform. Period. There are never two places involved.
> Therefore I
> would have to specify the target platform in the RMAP using an eclipse
> import reader. Unfortunately, it seems that an eclipse installation or an
> unzipped delta pack cannot be used by the import reader, because they do
> not represent a p2 metadata repository.
>
The trick is to define them as a target platform prior to resolution. They are both in 'runnable
state', i.e. features etc. has been unpacked, and as such, you can directly appoint them as a target
platform directory in a target platform definition.
Regards,
Thomas Hallgren
|
|
|
|
Powered by
FUDForum. Page generated in 0.03110 seconds
] 
Search
Help
Register
Login
Home
