GitLab

See go/faster-javac for a detailed analysis of why turbine was chosen to
speed up our java builds. The current implementation is an initial
implementation that can be further optimized. It already gives savings
of 10-30% of our java incremental builds depending on the target in
question (see go/project-turbo for timing details).

This implementation does remove some of the benefits of the previous way
that java_library_impl was implemented, namely when a target's
.interface.jar file changes, it no longer short-circuits dependent
targets from being built since we now use group(target_name). This is
arguably more correct as prebuilts depend on javac jars directly. In the
future (after the androidx migration and other build improvements) this
can be improved by having prebuilts depend on the interface jars
instead.

Bug: 1056299
Change-Id: I49d18604203749cf447ce77c54a8f8469c83eae1
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2105415
Commit-Queue: Peter Wen <wnwen@chromium.org>
Auto-Submit: Peter Wen <wnwen@chromium.org>
Reviewed-by: Andrew Grieve <agrieve@chromium.org>
Cr-Commit-Position: refs/heads/master@{#750651}