nonFinal, do we still need it?

[arjantijms]

The plug-in uses a convention for something called "nonFinal", which I think didn't carry over to Jakarta:

e.g.

//  OSGi Bundle-SymbolicName:   ${API_PACKAGE}-api
//  OSGi bundle specversion:    ${SPEC_VERSION}.99.b${SPEC_BUILD}
//  OSGi Bundle-Version:        ${SPEC_VERSION}.99.b${SPEC_BUILD}
//  jar Extension-Name:         ${API_PACKAGE}
//  jar Specification-Version:  ${SPEC_VERSION}.99.${SPEC_BUILD}
//  jar Implementation-Version: ${NEW_SPEC_VERSION}-b${SPEC_BUILD}

Do we need to retain this?

@bshannon @romain-grecourt

[romain-grecourt]

The versioning rules still mentions non final.

[arjantijms]

The versioning rules still mentions non final.

You're right, good find ;) Though is this "officially" adopted? We may also need an update here anyway for the -SNAPSHOT and -RCx cases.

One small thing, the option is now called "nonFinal", which means we often have to say "nonFinal" = false. Double negatives are in general hard for people to get right at a glance.

I know there's some history here, but going forward, can't we just call it final with values true and false?

[bshannon]

I think the intent was that final was the default if nothing was said so you would
only ever say nonFinal=true.

[romain-grecourt]

I do agree that double negations are hard to understand. A rename to final seems like a good future improvement.