feat: optimize py_binary start-up times: build venv with symlink actions instead of during py_binary bootstrap phase by gregmagolan · Pull Request #944 · aspect-build/rules_py

gregmagolan · 2026-04-23T23:05:52Z

No description provided.

aspect-workflows · 2026-04-23T23:31:22Z

Bazel 8 (Test)

Buildkite build #3274 is running...

Bazel 9 (Test)

Buildkite build #3274 is running...

Bazel 8 (Test)

e2e

⚠️ Buildkite build #3274 is currently failing.

Failing tests (1)

//cases/interpreter-features-836:test [k8-fastbuild]🔗

💡 To reproduce the test failures, run

bazel test //cases/interpreter-features-836:test

Bazel 9 (Test)

e2e

⚠️ Buildkite build #3274 is currently failing.

Failing tests (1)

//cases/interpreter-features-836:test [k8-fastbuild]🔗

💡 To reproduce the test failures, run

bazel test //cases/interpreter-features-836:test

Bazel 8 (Test)

examples/uv_pip_compile

All tests were cache hits

1 test (100.0%) was fully cached saving 444ms.

Buildifier

Buildifier managed files require formatting

--- ./py/private/py_image_layer.bzl	2026-04-24 06:43:14.753998545 +0000
+++ /tmp/buildifier-tmp-1944581848	2026-04-24 06:45:08.928799435 +0000
@@ -173,6 +173,7 @@
         srcs = [binary],
         **kwargs
     )
+
     # Rewrites `type=file content=` entries whose content path is
     # actually a symlink into proper `type=link` entries — otherwise
     # bsdtar follows the symlinks while building the archive, sees a
--- ./py/private/mtree_preserve_symlinks.bzl	2026-04-24 06:43:14.753998545 +0000
+++ /tmp/buildifier-tmp-3486313903	2026-04-24 06:45:08.952799183 +0000
@@ -32,6 +32,7 @@
         assignments.append(("owner", ctx.attr.owner))
     if ctx.attr.group:
         assignments.append(("group", ctx.attr.group))
+
     # Retained for contract compatibility with the upstream awk — the
     # plain-readlink-first branch in our script rarely consults it, but
     # absolute-symlink cases still use `resolved_path ~ bin_dir` for the

💡 Run the following to apply the suggested formatting fixes

bazel run //:buildifier

jbedard · 2026-04-24T02:48:38Z

        default = "error",
        values = ["error", "warning", "ignore"],
    ),
-    "interpreter_options": attr.string_list(


This just moved?

jbedard · 2026-04-24T02:49:21Z

+    # PyWheelsInfo (hand-written py_unpacked_wheel without `top_levels`, or
+    # non-wheel py_library deps) falls through to the .pth path below.
+    wheels_depset = _py_library.make_wheels_depset(ctx)
+    wheels = wheels_depset.to_list()


Did you think about avoiding the to_list() calls?

jbedard · 2026-04-24T02:54:51Z

    test_attrs = _test_attrs,
    toolchains = [
        PY_TOOLCHAIN,
-        VENV_TOOLCHAIN,


Does this mean the entire toolchain can be deleted?

…stead of during py_binary bootstrap phase

@gawk

@gawk is failing (05:40:22) ERROR: /mnt/ephemeral/output/rules_py/__main__/external/gawk+/BUILD.bazel:58:11: Compiling support/regex.c [for tool] failed: (Exit 1): clang failed: error executing CppCompile command (from target @@gawk+//:support) external/llvm++_repo_rules+llvm-toolchain-minimal-22.1.3-linux-amd64/bin/clang -target x86_64-linux-gnu -no-canonical-prefixes -Wno-builtin-macro-redefined '-D__DATE__=redacted' ... (remaining 70 arguments skipped) Use --sandbox_debug to see verbose messages from the sandbox and retain the sandbox build root for debugging In file included from external/gawk+/support/regex.c:71: external/gawk+/support/regexec.c:188:36: error: call to undeclared function '_REGEX_NELTS'; ISO C99 and later do not support implicit function declarations [-Wimplicit-function-declaration] 188 | size_t nmatch, regmatch_t pmatch[_REGEX_NELTS (nmatch)], int eflags) | ^ 1 error generated.

gregmagolan force-pushed the optimized_py_binary branch from 5de72ac to cc215c4 Compare April 23, 2026 23:31

gregmagolan force-pushed the optimized_py_binary branch 4 times, most recently from 51aff4f to 6a61c17 Compare April 24, 2026 01:27