Spotbugs: release-3.1 contains unexpected changes introduced by #832

[KengoTODA]

The base branch of #832 was unexpectedly release-3.1, that should be master according to its PR subject. We need to exclude 4bd6e43146e0739437d76f21298420efc5c4a4ad from release-3.1.

Here is current my idea about how to fix this problem:

git checkout 554d102c963fd8b318339ea046bb78385385cf8d
git branch -D release-3.1
git checkout -b release-3.1
git cherry-pick 8cd5f90266d2706a72176858ff0803586bd47fc5..0d1b933dc92a398ebf0cbbac7dca97585b577303
git push origin +HEAD

You can see the new branch created by this method at GitHub.

[KengoTODA]

@spotbugs/core-devs sorry for my careless, I made this problem when I propose & merge that PR.

This fix needs force-push so I want to ask some of you to confirm that my proposal is right to follow. Please review this process and the fixed-release-3.1 branch which will be new release-3.1 branch after this fix.

[henrik242]

Why do you need a force push? Just fix it properly by reverting the offending commit(s).

[KengoTODA]

I prefer force-push to make Git history intuitive. It helps us to ease tracing change by git blame etc.

I've prepared the reverted branch as reference, but I still prefer force-push solution...

[henrik242]

Making mistakes is a natural thing in development. Force pushing should be avoided at all costs, since it f*cks with the developers local trees. Also, they might wonder why their local tree failed if they haven't seen this issue. With a force push they would never know why.

Your reverted branch gets my +1, the force push doesn't :)

[iloveeclipse]

I haven't looked into details yet, but were there breaking changes merged? Do we actually need to change anything?

I'm not against force push, if there would be a quick note/recipe how to fix the cloned repos.
Revert will probably cause too many "false positive" results in git blame.

[KengoTODA]

OK I will propose a PR to merge a commit to revert. I've confirmed that its contents (files) is totally same with my branch that does not contains that PR. Thank you guys!