From aad2eb142c1ddd360b37381cdd3bfec269b57310 Mon Sep 17 00:00:00 2001 From: Thomas Schmelzer Date: Sat, 22 Aug 2026 22:33:50 +0400 Subject: [PATCH] chore: drop the exclude entries for the retired mutation/fuzzing workflows MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit rhiza v1.5.0 retired `.github/workflows/rhiza_fuzzing.yml` (#1568) and `.github/workflows/rhiza_mutation.yml` (#1572), and stopped offering mutation testing in the ecosystem at all (#1492). No bundle under `bundles/` ships either path now, so these entries decline a file the sync can no longer deliver — they describe a gate that no longer exists on either side. Nothing about CI changes: neither workflow is synced with or without the entries. Any copy still sitting in `.github/workflows/` stays there, because the orphan sweep only considers paths recorded in `template.lock`'s `files:` list and these are absent from it. Co-Authored-By: Claude Opus 5 (1M context) --- .rhiza/template.yml | 31 ------------------------------- 1 file changed, 31 deletions(-) diff --git a/.rhiza/template.yml b/.rhiza/template.yml index 4e0545c..f4b591c 100644 --- a/.rhiza/template.yml +++ b/.rhiza/template.yml @@ -21,37 +21,6 @@ exclude: # "Copyright (c) 2025 Jebel Quant Research", which is not this repo's holder. - LICENSE - # The fuzzing workflow, which needs FUZZING_ENABLED plus a `.clusterfuzzlite/` config this repo - # does not have. Its gate sits *inside* the reusable workflow rather than in the caller, so it - # starts on every pull request and every push to main in order to skip -- a green check whose - # only content is that it had nothing to do. - - .github/workflows/rhiza_fuzzing.yml - - # The mutation stub, for the same reason plus a worse one. This entry previously carried a note - # saying the caller was deliberately kept, because a MUTATION_ENABLED variable set to 'false' - # meant someone had configured the gate on purpose and flipping it to 'true' should be all it - # took. That premise was wrong: `gh api repos/Jebel-Quant/quadprog/actions/variables` returns - # `total_count: 0`, so the variable does not exist and never did. - # - # With nothing to flip, the stub's `if: vars.MUTATION_ENABLED == 'true'` can only skip -- and it - # triggers on `pull_request:`, so every PR starts a run in order to do so. Unlike the fuzzing - # caller the gate is at least in this file rather than inside the reusable workflow, but the - # visible result is identical: a green check whose only content is that it had nothing to do. - # - # It could not be turned on as it stands either. The reusable workflow fails the job if *any* - # mutant survives, and fourteen of this package's are provably equivalent, so a 100% gate can - # never go green here. - # - # Mutation testing is therefore not in CI in any form: the repo-owned weekly workflow that used - # to hold the middle ground -- scheduled rather than per-PR, gated on a baseline rather than on - # zero -- has been removed too, along with docs/development/MUTATION.md, which recorded the - # survivor baseline and which mutants are provably equivalent. Nor is there a working local - # route: rhiza-task's `mutation` task drives the mutmut 2.x CLI (jebel-quant/rhiza#1492), and the - # override that corrected that lived in the hand-written Makefile this repo replaced with the - # task-runner shim in v1.4. Recovering any of it starts with `git log --diff-filter=D` over the - # doc path above. - - .github/workflows/rhiza_mutation.yml - # The template's PAT_TOKEN and release-secret walkthrough. It documents secrets a consumer # configures once in the platform UI, addressed to whoever set the repo up rather than to anyone # reading the tree. Upstream is where it belongs, and where it stays.