|Re: Virgo over-constraing version numbers? [message #648778 is a reply to message #648696]
||Fri, 14 January 2011 13:52
| Glyn Normington
Registered: July 2009
Before we go any further, I'd like to know where these Require-Bundle headers are located so I can interpret the diagnostics more easily. We don't recommend using Require-Bundle because of the numerous semantic rough edges (enumerated in the OSGi spec) and so we don't use Require-Bundle in the Enterprise Bundle Repository (AFAIK) and bundlor won't generate Require-Bundle unless you explicitly code that header in the template.|
The funny thing is, the diagnostics give the impression that the org.apache.commons.cli bundle has Require-Bundle headers and I have no idea why such headers would have been placed in Apache bundles.
That said, I am not aware of any particular limitations in Virgo's handling of Require-Bundle. Apart from any scoping which Virgo might apply (due to a PAR or a scope plan) or provisioning of missing dependencies, the header should be passed through pretty much as-is for Equinox to resolve.
Anyway, apart from that, you should probably sort out:
A bundle could not be resolved because the required execution enviroment did not match the runtime environment. In bundle <org.apache.commons.lang_2.4.0.v201005080502>
because that looks like a hard error in your environment and may be triggering the other resolution failures.
Powered by FUDForum
. Page generated in 0.03738 seconds