[Bazel] Support for a custom prebuilt emscripten cache

[DoDoENT]

This adds support for using a prebuilt cache as an archive instead of building it from scratch every time.

This solves multiple problems:

• build speed on CI machines that have cold cache
  • instead of building the secondary cache every time, it can simply download it
• effective support for multiple caches (Bazel support for multiple emscripten caches #1581)
  • emscripten supports having multiple caches on its own, but it builds thinlto, lto, and other cache combinations on demand. This doesn't work in the Bazel world, where the cache needs to be frozen.
  • the current solution allowed building a secondary cache, but only with a single configuration
  • this allows the use of any archive that contain any emscripten cache
  • this PR does not handle how that cache is built - you can build it with embuilder, zip it and serve it from your server or whatever you want
• secondary cache hermeticity problem
  • the current solution for secondary cache generated emscripten_config file that contained full path to the cache.
  • this was not hermetic as this path could be different on different machines
  • this PR calculates the cache path by using the same trick with environment variables, ensuring that emscripten_config file is always same, regardless of the machine
  • additionally, it ensures that the cache is provided as an input to the toolchain so that Bazel can correctly add it to the sandbox environment

After applying this patch to our internal codebase, the regular builds of WASM code went down from 12-15 minutes to cca 30 seconds, regardless on which machine it runs, as long as remote cache is warm.

[sbc100]

@walkingeyerobot can you recommend anyone else who might be good to review this?

[DoDoENT]

@walkingeyerobot can you recommend anyone else who might be good to review this?

Since @walkingeyerobot didn't respond for a very long time, I added @mmorel-35 as a reviewer for this. He reviewed other Bazel-related emsdk PRs, so I'm confident he can take a look at this one as well.

We've been using this patch for over 5 months now in our company and haven't found any issues with it so far.

[nickschaap]

This is incredibly useful. We added this patch to our usage as well.

[sbc100]

I don't really feel qualified to review this. Maybe @mmorel-35 can chime in?

[Copilot]

Pull request overview

Adds support for supplying Emscripten’s frozen cache as a prebuilt archive (downloaded during repository setup) and makes the cache available as an explicit toolchain input to improve hermeticity and CI performance.

Changes:

• Extend emscripten_cache module extension to optionally download/extract a prebuilt cache archive and wire it into emscripten_config.
• Add a builtin_cache filegroup and an emscripten_cache target so the toolchain always carries cache files into the sandbox (builtin or secondary/prebuilt).
• Document prebuilt_cache usage in the Bazel README.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

File	Description
bazel/remote_emscripten_repository.bzl	Adds cache filegroup target to toolchain “common files” inputs.
bazel/emscripten_cache.bzl	Adds prebuilt_cache support and switches cache path computation to be machine-independent.
bazel/emscripten_build_file.bzl	Exposes the builtin Emscripten cache as a filegroup for toolchain inputs.
bazel/README.md	Documents configuring a prebuilt cache archive via Bzlmod.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.