RFR: update-sync labels should be applied to all-bar-one records

[Rob McKenna]

On Mon, 28 Sep 2020 23:24:32 GMT, Kevin Rushforth <kcr at openjdk.org> wrote:

>> This modification fixes the situation where a backport record may be a member
>> of two distinct streams and has its hgupdate-sync label removed as a result.
>> 
>> E.g. with backport records for 14, 15, 14.0.1, 15.0.1 no label would be
>> applied to 15. The current logic gives precedence to the fact that it is the
>> first release in the 15-updates stream and ignores the fact that it is not
>> the first release in the feature stream. The label added by the feature stream
>> is removed when evaluating the 15-updates stream. This should not be the case.
>
> The fix looks good to me. One question about the test: is the `assertNotLabeled` method really needed? It looks like
> `assertLabeled` will fail unless the set of issues with the labels is exactly as expected (i.e., it will fail if it
> finds an issue that should have the sync label but doesn't or if if find an issue that should not have the sync label
> but does), unless I'm missing something.

I think I like the lack of ambiguity in the name. I vote to leave it in, but if Robin objects I can remove it.

-------------

PR: https://git.openjdk.java.net/skara/pull/848