ICU-23091 Add Java source formatting configs and automated checks

[echeran]

Add automated source formatting so that we don't have to worry about it again. In the future, the diffs that are visible in PRs will no longer have noise (extra diff lines) due to formatting unrelated to the code issues that are the focus of the PR.

Checklist

• Required: Issue filed: ICU-23091
• Required: The PR title must be prefixed with a JIRA Issue number. Example: "ICU-1234 Fix xyz"
• Required: Each commit message must be prefixed with a JIRA Issue number. Example: "ICU-1234 Fix xyz"
• Issue accepted (done by Technical Committee after discussion)
• Tests included, if applicable
• API docs and/or User Guide docs changed or added, if applicable

ALLOW_MANY_COMMITS=true

[jira-pull-request-webhook]

Notice: the branch changed across the force-push!

• .github/workflows/icu4j.yml is different

View Diff Across Force-Push

~ Your Friendly Jira-GitHub PR Checker Bot

[echeran]

For review purposes and consideration:

• There are 2 commits in the PR: one has the key config changes and doc updates, and the other has the changes to the remaining codebase Java source files
• Let's decouple the work of reviewing/approving from the action of actually merging the PR, and not assume that we do these things at the same time for this PR
• When this gets merged, then any concurrent open PRs that touch Java files will need to update themselves with the latest on main to pick up the new source formatting. That will cause churn. We can choose the right time to merge. Of course, likewise, this PR will need to pick up the latest from main for any Java changes between now and when we actually merge

[markusicu]

@yumaoka @richgillam FYI

[markusicu]

• thanks!
• what about the 11 failing ("cancelled") checks?
• config looks plausible
• code reformatting changes... I know not to quibble with the formatter...
• we will need to talk about when to merge this
• we might want to push everyone (mea culpa etc.) to move on pending pull requests and reviews, so that there are fewer pending when this goes in

[mihnita]

LGTM
Thank you.

[echeran]

• what about the 11 failing ("cancelled") checks?

I retried them and all but one succeeded. The ones that passed after retrying were all from the ICU4C CI workflow, and given that this PR here is only about source code formatting for Java, it's safe to conclude that those ICU4C check failures during the first attempt are unrelated. The other check that failed is the "meta check" job called "Wait for Required Checks", and it fails because the single-commit check fails. For now, I am leaving the two commits in the PR to remain separate in order to make it easier to understand when reviewing.

[markusicu]

The other check that failed is the "meta check" job called "Wait for Required Checks", and it fails because the single-commit check fails. For now, I am leaving the two commits in the PR to remain separate in order to make it easier to understand when reviewing.

We should add the magic incantation that allows us to rebase+merge the two commits.

[echeran]

We should add the magic incantation that allows us to rebase+merge the two commits.

Done.

[echeran]

@markusicu Ready again when you are. I don't see any PRs that simultaneously: 1) touch ICU4J 2) are ready to merge 3) are recent enough to consider. The only possible exception is maybe #3468, but that's very small, enough so to not hold up for.

[markusicu]

@markusicu Ready again when you are.

Ok... :-)

I reminded myself that the TC is on board with this now:

• 2025-08-21: When do we start auto-formatting the Java source code? Mihai proposes that we do this right after the release (or else ASAP) / after merging maint-78 to main [...] Agreed
• 2025-07-17: Consensus: Let’s do this before 78 release front loading – soon after aug07
• 2025-06-12: Let’s decide when a good time to merge is. Preferably a time when is least disruptive to us & PR contributors

I don't see any PRs that simultaneously: 1) touch ICU4J 2) are ready to merge 3) are recent enough to consider.

+1

The only possible exception is maybe #3468, but that's very small, enough so to not hold up for.

And it already has a merge conflict.

@richgillam @mihnita @yumaoka FYI

[markusicu]

@echeran if you reversed the commits, it would be easier to keep the GitHub UI alive to get to the approve button...