Release candidates should not steal the maven index from GA releases

[mbien]

lets not move the index from NB 24 -> 25-rcN when testing new releases. This would only cause NB 24 to download it again which is unnecessary.

(if this happened to you you can simply move it back to the 24 cache. Setting -J-Dmaven.indexing.diag.release=0 will also prevent it)

targets delivery

[neilcsmith-net]

Looks good. Slightly concerned at proliferation of code that relies on specifics of our versioning scheme. The next time we need that, something to possibly considering abstracting into Places and/or a way to mark prereleases with a file similar to lastUsedVersion.

[mbien]

Looks good. Slightly concerned at proliferation of code that relies on specifics of our versioning scheme.

yes, we have several instances were it would be useful to query the NB version etc (nb version, javac versions #3274).

But this here is just a "best effort" mechanism and version is just one attribute to probe for, the other are time stamps etc. If this fails it fails quietly and is no show stopper.

The next time we need that, something to possibly considering abstracting into Places..

yep!

[ebarboni]

LGTM