[TrapsterDK]

gazelle/julia/BUILD.bazel loads julia_binary/julia_test from @rules_julia directly, so consumers without patches/rules_julia:package_name.patch no longer fail analysis of stdlib_dump_upstream.

Part of #11418. Remaining: rules_cuda, rules_python, rules_go, rules_julia upstream patches/releases, protobuf consumer test.

Tests: //gazelle/julia:stdlib_test passes locally. Nested-Bazel and consumer tests not run.
Evidence: windows_x86_64 //tools/testing/hermetic:cpp_rust run 43660885365 green (dx_generate_check ok, Kotlin coverage green, qualified). Earlier run 43537493312 failed on a duplicate dx_licenses tag, since fixed. This branch also carries 12780 (merged) and 12703.

🤖 Generated with Claude Code

https://claude.ai/code/session_01Xut2hws3L7Tg4oUvpSvuYD

[TrapsterDK]

Windows evidence for //tools/testing/hermetic:cpp_rust, run 43541761804 (test_timeout=3000): fails past analysis now. The consumer's rules_rust process_wrapper link fails with rust-lld 'could not open kernel32.lib' because the unpatched llvm module sets LIB=hermetic_llvm_empty_lib (patches/llvm/msvc_link_lib_env.patch is root-only). Needs a consumer-visible llvm patch (#11416). Stays draft.

[TrapsterDK]

Evidence run 43552270331 (windows_x86_64, //tools/testing/hermetic:cpp_rust, test_timeout=3000) now links and passes build/fetch/leak probes; fails at no_refetch: a restarted server refetched repositories (winsdk, host_platform, rules_java toolchains, rules_dx+). Previous run 43551159922 hit the 600s default timeout.

[TrapsterDK]

Root cause of the no_refetch failure in run 43552270331: every refetched repo is a local = True repository rule (bazel_tools winsdk_configure, platforms host_platform, rules_java local_jdk / compatibility_proxy / remote*_toolchain_config_repo, rules_android androidsdk, rules_cuda + rules_dx cuda configure, rules_py uv host) or the local_path_override of rules_dx itself (@@rules_dx+). Bazel documents that local = True repos are refetched on every server restart, so the probe added in #11476 (tools/testing/hermetic/src/lib.rs, no_refetch) cannot pass on any platform as written; it is not Windows-specific and not in anything this PR touches. Non-local repos (llvm, rules_rust, rules_python, ...) were not refetched.

Main never reaches this probe: main's linux, macOS and Windows hermetic jobs for //tools/testing/hermetic:cpp_rust fail earlier (fetch / hermetic_test) since #11476 landed (e.g. CI runs 43414090265, 43208363961).

Deciding how the guard should treat local repos (filter local_path_override modules as the nested harness does, and/or upstream local rules) is a guard change outside this PR's scope, so this PR stays draft pending that fix on main.

🤖 Generated with Claude Code

https://claude.ai/code/session_01Xut2hws3L7Tg4oUvpSvuYD

[TrapsterDK]

Rebased on main (clean, 12641 tests pass locally). Combo branch combo-11418-11443 (12641+12723+12703), //tools/testing/hermetic:cpp_rust:

• linux_x86_64 run 43568707405: fails at dx_generate_check, env: 'bash': No such file or directory (gazelle runner #!/usr/bin/env bash under the coreutils-only PATH from hermetic: qualification PATH is the pinned coreutils applet directory (part of #10673) #11592). Main issue, not in this chain; main never reached it because fetch failed until rules_py: uv host repo reports glibc 2.28 without running ldd (part of #12482) #12485.
• windows_x86_64 run 43568709411: passed dx_generate_check, failed at refetch_rules (command line over 1024 chars). Fixed on 12703 (batched show_repo); re-run 43571299476 in progress.
  Not merged: linux evidence cannot go green until the bash PATH issue is fixed on main.

🤖 Generated with Claude Code

https://claude.ai/code/session_01Xut2hws3L7Tg4oUvpSvuYD

[TrapsterDK]

Status of the chain (12641 -> 12723 -> 12703), combo branch combo-11418-11443 rebased on main (windows_support conflict: reproducible_extensions.patch now applies after main's pins.patch and keys off the pinned values; lock regenerated with Bazel) plus #12753.

• dx: generate/generate_check run gazelle without a bash runner (part of #11443) #12753 (merged): dx:generate/dx:generate_check are native_binary wrappers of the gazelle binary, so no #!/usr/bin/env bash under the coreutils-only PATH.
• linux_x86_64 run 43575015910: passes coverage and reaches dx_generate_check, which now fails on Gazelle drift (size = "small" on kotlin_test in kotlin/tests/fixtures/kotlin-app). Blocked on main: generate drift and generated-doc tests fail #12372 (PR 12739, which covers cli/ but not grades/).
• windows_x86_64 run 43571299476: passes no_refetch, fails at coverage (kotlin_test InvalidTestClassError on BazelTestRunner). Consumers do not get patches/rules_kotlin/windows_coverage_test_suite.patch; blocked on a rules_kotlin release with Do not override bazel.test_suite in the Windows coverage launcher bazel-contrib/rules_kotlin#1829 (jvm: drop Windows coverage workarounds once rules_kotlin#1829 and bazel#32479 ship #9069).
  Not merged.

🤖 Generated with Claude Code

https://claude.ai/code/session_01Xut2hws3L7Tg4oUvpSvuYD

[TrapsterDK]

Rebased onto #12780 (fix-9069-kotlin-consumer-patch). Evidence run 43651406826 (cpp_rust, windows_x86_64) covers both. Merge #12780 first.

[TrapsterDK]

Run 43651406826 (Windows cpp_rust): every probe passed, including fetch_coverage. The run failed only at the no_refetch check: "a restarted server refetched repositories" (local_config_winsdk, host_platform, local_jdk, ...). #12703 fixes that check. I stacked #12703's 3 commits, plus a unit-test fix that writes the rules_julia patches in the_consumer_materializes_from_dx_init (that test also fails on main), onto fix-11418-julia-package-name. Then I dispatched run 43653857809. The coverage probe, which covers Kotlin, runs after no_refetch, so it is still unverified.