Github user ahgittin commented on the issue:
https://github.com/apache/brooklyn-server/pull/740
i've cleaned up several things, but multiple discussion threads remain:
> DefaultMaven2OsgiConverter different
for recommended versions it is the same, isn't it? that's the important
thing. for other things yes it will be surprising but someone making an OSGi
bundle relying on that converter's obscure behaviour is asking for trouble.
i'd rather optimize to be sensible for people doing foolish non-recommended
things in non-maven / non-OSGi cases, where version `1.2_SNAPSHOT` is probably
`1.2.0.SNAPSHOT` not `1.0.0.2_SNAPSHOT`.
> sanitize catalog item IDs filenames
again only an issue if using non-recommended IDs, and we don't rely on
those files very much now (esp once YAML BOMs are added as bundles), so content
to leave this as a bug.
> different ordering to OSGi
semver says qualifer makes something less than unqualified, e.g.
`1.0.0-rc1` < `1.0.0`, and i think that's what people expect now. also i think
they are surprising i the case of `-rc9` v `-rc10`.
our ordering is different to both; we should make this clear in the docs
but i think that's better than forcing people to understand the surprising
ordering in osgi.
the one place this could be an issue is if we've relied on OSGi to resolve
version ranges. but i think let's cross that bridge when we come to it, rather
than have `1.0.0-rc1` > `1.0.0` -- no?
> comparators shouldn't return 0
discussed above, but captured here for completeness
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
