Re: Resolving to micro version

2018-04-04 Thread Katsuya Tomioka
That explains. I had particularly hard time imagining the clause in the spec. (now I have an example.) Thanks all for help. -Katsuya On Wed, Apr 4, 2018 at 8:20 AM, Neil Bartlett wrote: > Thank you, now the answer is clear. > > You can see that the lower version of rbl-osgi (bundle ID 9) has z

Re: Resolving to micro version

2018-04-04 Thread Neil Bartlett
Thank you, now the answer is clear. You can see that the lower version of rbl-osgi (bundle ID 9) has zero active exports, and it is importing the package com.basistech.rbl.osgi from the higher version of itself. This is because rbl-osgi has both an import and an export of that package. When this h

Re: Resolving to micro version

2018-04-04 Thread Karl Pauls
Can you please share the manifests/headers of the bundles involved with us? regards, Karl On Wed, Apr 4, 2018 at 2:04 PM, Katsuya Tomioka wrote: > Thanks for the explanation. The narrow import range is intentional, I > manually set the import in bnd (via bnd-maven-pugin) to satisfy functional >

Re: Resolving to micro version

2018-04-04 Thread Katsuya Tomioka
Thanks for the explanation. The narrow import range is intentional, I manually set the import in bnd (via bnd-maven-pugin) to satisfy functional requirements; I need to consume the version specifically. The runtime currently happened to have both versions of the component. g! inspect cap osgi.wir

Re: Resolving to micro version

2018-04-03 Thread Neil Bartlett
What Ray said is absolutely true, but I don’t think it actually explains the problem. As far as I can tell, the range should actually catch the export from the "rbl-osgi (7.24.0.c592)” bundle. Katsuya-san: just to be sure, please type the following commands and send the output: inspect

Re: Resolving to micro version

2018-04-03 Thread James Carman
Yes, sorry, was on the road before. It uses semantic versioning to determine what matches and 7.24.1 is much less (99 less) than 7.24.100. It doesn’t compare them lexicographically. It treats each section as an integer. Are you using the maven bundle plugin? If not, you should be. It, for a m

Re: Resolving to micro version

2018-04-03 Thread Raymond Auge
I think that, as James was alluding to, your problem is in the package import range you stated in your original message: "[7.24.0,7.24.1)" Do you know how the import range came to be so narrow? Typically the range is something like "[7.24.0,7.25.0)" Perhaps the import policy was altered? - Ray

Re: Resolving to micro version

2018-04-03 Thread Katsuya Tomioka
Thanks for the reply. I hugely omitted details, I realized. This is the error from gogo with felix 5.6.10: ERROR: Bundle rex-osgi [13] Error starting file:/Users/kt/work/felix/ felix-framework-5.6.10/bundle/com.basistech.rex-je-rex-osgi-7.33.101.c59.2.jar (org.osgi.framework.BundleException: Una

Re: Resolving to micro version

2018-04-03 Thread James Carman
7.24.100 > 7.24.1 On Tue, Apr 3, 2018 at 5:17 PM Katsuya Tomioka wrote: > Felix users, > > I have same bundles (say "A") installed only different by micro versions, > 7.24.0 and 7.24.100. I have then another ("B") imports packages from "A" > such that: "[7.24.0,7.24.1)". This works fine if I ins

Re: Resolving to micro version

2018-04-03 Thread Neil Bartlett
> On 3 Apr 2018, at 22:17, Katsuya Tomioka wrote: > > Felix users, > > I have same bundles (say "A") installed only different by micro versions, > 7.24.0 and 7.24.100. I have then another ("B") imports packages from "A" > such that: "[7.24.0,7.24.1)". This works fine if I installed 7.24.0 only.

Resolving to micro version

2018-04-03 Thread Katsuya Tomioka
Felix users, I have same bundles (say "A") installed only different by micro versions, 7.24.0 and 7.24.100. I have then another ("B") imports packages from "A" such that: "[7.24.0,7.24.1)". This works fine if I installed 7.24.0 only. But soon as 7.24.100 installed, the bundle B fails to resolve wi