Stop submoduling re2 (#8935)
commitc6272e834690a8e370bcaccb6018f154cd48bc0e
authorFred Emmott <fe@fb.com>
Fri, 19 Nov 2021 23:13:01 +0000 (19 15:13 -0800)
committerFacebook GitHub Bot <facebook-github-bot@users.noreply.github.com>
Fri, 19 Nov 2021 23:14:53 +0000 (19 15:14 -0800)
tree080b4bda1b868d0aaa2be090af8f0261fe1897da
parentd6520899df0f007632fd5c8774604a64e52b7421
Stop submoduling re2 (#8935)

Summary:
- Use release tarball instead of stale submodule
- faster builds
- more consistent with other stuff
- always defines re2 target, conditionally defines bundled_re2 target
- resolve re2 build TODO in fbmysqlclient CMake, which required this split (fooBuild was the old name for bundled_foo targets) by making it depend on re2 (but not bundled_re2)
- clean up fbmysqlclient's cmake to get it up to speed with our current conventions while I'm here

The net effect is that:
- build of re2 should be faster, as we'll be fetching a source tarball instead of gitsubmodule - and in FB infra, it's cached locally
- we won't spend time fetching re2 sources at all if the system version is new enough
- fbmysqlclient will also build faster if system re2 does not exist or is too old: previously HHVM and fbmysqlclient would each build their own copy

Pull Request resolved: https://github.com/facebook/hhvm/pull/8935

Test Plan: *castle and built on AWS worker.

Reviewed By: shayne-fletcher

Differential Revision: D32537910

fbshipit-source-id: 6bcbc9eb7e69131dfe68528faddbf6eb65b470aa
.gitmodules
third-party/CMakeLists.txt
third-party/fb-mysql/CMakeLists.txt
third-party/re2/CMakeLists.txt
third-party/re2/src [deleted submodule]