Those are questions that are hard to answer.
My guess would be if you just have a few "pre-reqs" (such as
"just SDK and XML"), you'd be better off specifying those specific
features (not even categories). If you have a large number of complicated
pre-reqs, my quess is you'd be better off creating your repo during your
build ... all three of the main ones at use at Eclipse projects (that I
know about) have ways of "creating a repository" and various
methods of specifying required pre-reqs. Sounds like you have a situation
where there are lots of methods to accomplish what you need, and the hard
part is knowing which is easiest.
Good luck ... let us know what you learn!
Stephan Herrmann <stephan.herrmann@xxxxxxxxx>
10/08/2013 07:22 AM
b3 aggregator can't consume kepler repo?
Thanks, David, for comprehensive hints!
It seems the normal solution will be to specify kepler
as a validation repo rather than as a mapped repo. Good.
In some cases, however, I have the requirement to create
a self contained repo. For that scenario I need to find
a way to specify kepler as a mapped repo with a suitable
filter so I don't get any contradictory requirements.
It seems I have two options:
- positively list contents, e.g., via categories
- define exclusion rules for all non-IDE stuff
Would excluding all of RAP and all target platforms
already do the job in this latter direction, or is
more trial and error to be expected?
Would I need to browse the full meta data for max=0,
On 10/08/2013 03:34 AM, David M Williams wrote:
> That's not a very good error message, is it.
> But, if I had to guess (your description is pretty high level) I suspect
you are basically telling b3 aggregator to combine "all of
> Kepler" with your repo ... right so far? If so, then I
suspect you are running into the (intentional) "negative dependencies"
> are defined for some "runtime only features", that are there
to explicitly prevent some features from "being installed together"
> in this case, to prevent some "runtime only" features, say,
from RAP, from being installed into the IDE.
> As background on these topics, you might want to read
> or p2 mailing list,
> This "advanced technique" was introduced in Kepler. Previously,
in Juno, everyone used a "fake" A.PDE.Target.Platform requirement
> which only the aggregator knew about with respect to particular category.
With the negative requirements now in use, p2 knows about
> it too, and you can specify, for example, as RAP does, "install
this feature, or SWT, but never both together" (or, something like
> Now, I could guess at what you are trying to accomplish, and offer
some possible solutions, but then this reply would get really
> long. :)
> But, I'll give some hints ... the answer is probably in
> or, indirectly, in
> If not, you might try a "how do I ... " question on the
> From: "Stephan Herrmann" <stephan.herrmann@xxxxxxxxx>
> To: cross-project-issues-dev@xxxxxxxxxxx,
> Date: 10/07/2013 04:40 PM
> Subject: [cross-project-issues-dev] b3 aggregator can't consume kepler
> Sent by: cross-project-issues-dev-bounces@xxxxxxxxxxx
> I've had some fights with the b3 aggregator today and it seems to
boil down to the fact that the aggregator cannot consume the
> kepler repository, neither releases/kepler, nor one of the children
201306260900 or 201309270900.
> To witness: I have a toy aggregation (1 feature with one plugin) that
validates OK, but as soon as I add the kepler repo the
> aggregator dies with:
> org.eclipse.core.runtime.CoreException: Cannot complete the install
because some dependencies are not satisfiable
> at org.eclipse.b3.aggregator.engine.ValidationSetVerifier.run(ValidationSetVerifier.java:686)
> at org.eclipse.b3.aggregator.engine.Builder.runRepositoryVerifier(Builder.java:1704)
> at org.eclipse.b3.aggregator.engine.Builder.run(Builder.java:1635)
> at org.eclipse.core.internal.jobs.Worker.run(Worker.java:53)
> It's the same exception as in https://bugs.eclipse.org/bugs/show_bug.cgi?id=321756#c11
> Adding the juno SR2 repo doesn't have this effect (full juno repo
has a conflict in a modisco feature).
> Has anybody else seen this?
> Are those repos broken, or is it "just" b3?
> FWIW I tried 4.2 and 4.3 versions of b3.
> cross-project-issues-dev mailing list
> cross-project-issues-dev mailing list