Dne pátek 26. srpna 2022 13:29:54 CEST, Neil C Smith napsal(a): > On Sat, 20 Aug 2022 at 07:37, Jaroslav Tulach <[email protected]> wrote: > > Executive summary: If a "cleanup PR" has to touch `.sig` files, then it is > > not a cleanup! > > Totally agree! > > However, the interesting thing here is that if it wasn't for the > bitshift value changes showing up in > https://github.com/apache/netbeans/pull/4487 this might not have been > picked up. > > Shouldn't the constant value changes have also caused test failures in > the original cleanup PR?
Hmm. Good point. The explanation very likely is: we are running "binary compatibility" check. E.g. not what happens after re-compilation, but what happens during linkage time. `public static final int` constants are copied to the usage place by value at compilation time. E.g. removing such constant isn't binary incompatible! Just source incompatible! The bitshift made the change also binary incompatible as the previous values of the constants and the new values were different. Interesting behavior of sigtest. I wasn't aware of it. > Maybe in Michael's CI changes we could also look at failing PRs that > touch .sig files unless they have a particular label set on them (eg. > API change)? Labeling API changes by CI would be great! -jt > Best wishes, > > Neil > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > > For further information about the NetBeans mailing lists, visit: > https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected] For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
