| commit | 9999f3a9835df8dd8e8e0c82bc15fe46b69a6526 | [log] [tgz] |
|---|---|---|
| author | David 'Digit' Turner <digit@fuchsia.infra.roller.google.com> | Wed Apr 03 19:01:50 2024 +0000 |
| committer | Copybara-Service <copybara-worker@google.com> | Wed Apr 03 12:04:47 2024 -0700 |
| tree | bcd08f32ab49a614b78d350cc1282a854cb3c703 | |
| parent | 4c012aeb9569b8a529bc3ef358d69c45d75180b3 [diff] |
[roll] Roll fuchsia [build][bazel] Keep output_base when updating workspace. Before this CL, update_workspace.py would clear the entire Bazel output_base whenever the content of the Bazel workspace needed to be updated. This could take up to 27s on a powerful workstation. This CL makes the script complete in 0.6s instead! Cleaning the whole output_base is not necessary, as it can hold 3.4 GiB of build artifacts that can totally be reused for the next Bazel invocation, since Bazel doesn't use timestamp, only content hashes, to run build actions. This CL keeps all build artifacts in the output_base, but still clears its `external/` sub-directory which holds the content of external repositories. In Bazel, repository rules are not hermetic or reproducible, and making them incrementally correct is very difficult. As the Fuchsia build relies on more and more third-party repository rules, it is impossible to guarantee correctness. Nuking the external/ sub-directory forces all repository rules to re-run from scratch after an update, which is still considerably faster than recreating all build artifacts. NOTE: This CL does not address the fact that even a simple `git switch` will force a dozen successive Ninja -> Bazel invocations, even if the latter have nothing to do. This is a completely different issue, which will be addressed separately. Original-Reviewed-on: https://fuchsia-review.googlesource.com/c/fuchsia/+/1019832 Original-Revision: 57429a7fa52e3dba9bb0ac2a49bf0f302a582602 GitOrigin-RevId: 2669981b9f5bec2a29d361f3cb477dd208ac69a8 Change-Id: I27936ca9c54d8b7a8b8aec6fe97613680ee6d081
This repository contains Fuchsia's Global Integration manifest files.
All changes should be made to the internal version of this repository. Our infrastructure automatically updates this version when the internal one changes.
Currently all changes must be made by a Google employee. Non-Google employees wishing to make a change can ask for assistance via the IRC channel #fuchsia on Freenode.
First install Jiri.
Next run:
$ jiri init $ jiri import minimal https://fuchsia.googlesource.com/integration $ jiri update
Third party projects should have their own subdirectory in ./third_party.