Repository navigation
ESTATE: de-vendor embedded affinescript trees (bofj-kitt / proof-burrower / developer-ecosystem) #230
Description
Is your feature request related to a problem? Please describe.
The estate dialect-conformance audit found 3 repos that vendor a full copy of
the affinescript tree rather than depending on it:
bofj-kitt—affinescript/{conformance,docs,examples,stdlib,test,test_if.affine}(~218.affine)proof-burrower— same shape (~221.affine)developer-ecosystem—affinescript-ecosystem/{affinescript-deno-test,rattlescript}(~208.affine)
Their large "drift" counts in the audit are the vendored affinescript
negative/test corpus, not independent consumer drift — i.e. phantom drift
that tracks whatever upstream snapshot was copied. Risk: the vendored copies
silently rot vs upstream main (stale stdlib/grammar/conformance), and they
inflate every estate-wide .affine scan.
Describe the solution you'd like
De-vendor: replace each embedded affinescript tree with a pinned dependency
(submodule / package ref / documented snapshot with a refresh job), so there is
one source of truth and no phantom drift.
Describe alternatives considered
- Leave as-is — rejected: perpetual stale-snapshot risk + scan noise.
- Fold into the dialect-conformance campaign — rejected: distinct problem class
(supply-chain / vendoring policy, not dialect drift). Tracked separately here.
Additional context
DEFERRED — explicitly out of scope for the dialect-conformance campaign
(#228/#229); filed so it is not silently forgotten (user decision 2026-05-18).
Not blocked by #228. Audit harness + evidence:
/home/hyperpolymath/dev/repos/.affine-audit/ (results.tsv rows for
bofj-kitt / proof-burrower / developer-ecosystem).
Activity
hyperpolymath commentedon May 19, 2026
Cross-link from the #229 closure (2026-05-19) — #230 is the home for this.
The #229 post-port truthing (
docs/RESCRIPT-ELIMINATION.adoc§"#229closure status") found that every ReScript-surface file the audit
flagged in
developer-ecosystem,proof-burrower, andbofj-kittis avendored copy of the affinescript repo itself —
affinescript/stdlib/effects.affine,affinescript/tests/conformance/run_all.affine,affinescript/docs/guides/warmup/02_ownership.affine(dev-ecosystem nestsit deeper under
affinescript-ecosystem/rattlescript/affinescript/…). Themutablematches were comments ("// State effect - mutable references","// ── mut: mutable borrow ──"), i.e. scanner false positives — the same
discipline as the affinescript-own-corpus exclusion and the
panllFP.Consequence:
reclassified OUT of ESTATE: total ReScript-surface elimination from all .affine (committed goal; blocked-on #228) #229 permanently (recorded in
RESCRIPT-ELIMINATION.adoc).affinescript trees — which is exactly this issue (ESTATE: de-vendor embedded affinescript trees (bofj-kitt / proof-burrower / developer-ecosystem) #230), not ESTATE: total ReScript-surface elimination from all .affine (committed goal; blocked-on #228) #229.
ESTATE: total ReScript-surface elimination from all .affine (committed goal; blocked-on #228) #229 will not re-list them.
No action change to #230's scope; this just makes it the explicit,
authoritative home so the overlap is not re-litigated under #229.
Refs #229.
hyperpolymath commentedon May 27, 2026
Cross-reference (triage 2026-05-27): scope-disambiguated against #229 and #392 — these three are distinct, not consolidatable.
.affinefiles in independent consumer repos (burble, idaptik-dlc-vm, git-scripts, …). Blocked-on LANG: type/effect grammar has no module-qualified path —Pkg.Type/Pkg.Effectunrepresentable (estate-wide port blocker; ADR-014) #228. Per this issue's existing comment, the 3 vendored-tree repos were reclassified OUT of ESTATE: total ReScript-surface elimination from all .affine (committed goal; blocked-on #228) #229.Pkg.Type/Pkg.Effectunrepresentable (estate-wide port blocker; ADR-014) #228.affinescript-*subtrees to their own repos (or to developer-ecosystem/) #392 — extraction: push five misplaced subtrees (affinescript-vite/,affinescript-dom/,affinescript-pixijs/,affinescript-tea/,affinescriptiser/) OUT of the affinescript repo itself into their proper homes. Reverts the 2026-04-22 consolidation mistake.#230 pulls copies INTO upstream alignment; #392 pushes subtrees OUT of upstream. Inverse directions, disjoint target sets, distinct problem classes. Consolidation is not safe.
hyperpolymath commentedon Sep 19, 2026
Status measured today (2026-09-19) — one of the three is done, one is ambiguous and needs your ruling before anything is deleted.
bofj-kittaffinescript/path exists onmainproof-burroweraffinescript/developer-ecosystemaffinescript-ecosystem/holdsaffinescript/,affinescript-vite/,affinescriptiser/,rattlescript/,affinescript-deno-test/— 1,701 entriesThe ambiguity is with #392, and it is not mine to resolve. #392 asks for the binding subtrees (
affinescript-vite/,affinescript-dom/,affinescript-pixijs/,affinescript-tea/,affinescriptiser/) to be moved out of the affinescript repo and into developer-ecosystem — i.e. developer-ecosystem is the intended destination for those trees. But theaffinescript/copy sitting insideaffinescript-ecosystem/is still a vendored duplicate of the main repo, which is what this issue is about.So:
proof-burroweris unambiguously actionable;developer-ecosystemneeds #230 and #392 reconciled first — either the vendoredaffinescript/is removed and only the binding subtrees stay, or the two issues agree a different shape. I have not deleted anything there.