From d73a2caf0c58cc7cd6fa22e6101c1e3da2522fe6 Mon Sep 17 00:00:00 2001 From: Flotapponnier <160007691+Flotapponnier@users.noreply.github.com> Date: Wed, 28 Oct 2026 14:26:59 +0100 Subject: [PATCH] main's provider-set hash is a day and two merges out of date The guard has been red on main since 2026-10-27 and it was right to be. Two merges moved a bench's provider set and neither touched a cache key or this hash: #3311 stopped scoring Mobula on Solana in token-quote-coverage, and #3312 added Perpex to hyperliquid-frontends. Both changed only benchmarks/*.yml, which is precisely the shape this test was written to catch, since a provider set can move without anyone editing the code that caches it. Red CI on main greys out every PR opened against it, so this unblocks the branch as much as it records anything. The four keys are NOT bumped, and that is a decision about timing rather than a weakening of the rule. A bump invalidates entries written before the change. By today the cache has already turned over on its own, which I measured instead of assuming: production serves 106 providers on hyperliquid-frontends with Perpex among them, and /products/perpex answers 200 carrying its hyperliquid-frontends appearance, which is the appearance providers-vP exists to protect. Bumping now would flush healthy entries and reveal nothing. Caught on the day, the keys would have had to move, and for the next change they still do. The comment above the hash says all of that, so the next reader does not take this as permission to skip a bump. Recorded hash moves f355324b6640e7e4 to b767efc1bf40bf99 against 1841 providers. bun test 786 pass 0 fail, tsc clean, eslint clean. --- src/lib/spec-provider-set.test.ts | 28 +++++++++++++++++++--------- 1 file changed, 19 insertions(+), 9 deletions(-) diff --git a/src/lib/spec-provider-set.test.ts b/src/lib/spec-provider-set.test.ts index 6934f6dc3..487665135 100644 --- a/src/lib/spec-provider-set.test.ts +++ b/src/lib/spec-provider-set.test.ts @@ -53,15 +53,25 @@ describe("the provider set behind the cache keys", () => { const hash = createHash("sha256").update(sets.join("\n")).digest("hex").slice(0, 16); const count = sets.reduce((n, s) => n + s.split(":")[1].split(",").length, 0); - // Recorded on 2026-10-24 with bench-unfiltered-v99, all-benchmarks-v92, - // bench-filters-v40, providers-v19. + // Recorded on 2026-10-28 on main, with bench-unfiltered-v102, + // all-benchmarks-v95, bench-filters-v43, providers-v21, none of which + // moved. Read the next paragraph before copying that as a precedent. // - // The change was a DELETION, bench 306, which moves this hash the same - // way an addition does but for the opposite reason: a cached aggregate - // written before it would still list a bench that no longer exists. - // providers-v did NOT move, because both of its providers (pump-bridge, - // relay) are declared by benches that remain, so no product page gains - // or loses an appearance. + // Two merges moved the set and neither bumped a key: #3311 stopped + // scoring Mobula on Solana in token-quote-coverage, and #3312 added + // Perpex to hyperliquid-frontends. Both touched only benchmarks/*.yml, + // so this guard went red on main on 2026-10-27 and stayed red, which is + // exactly what it exists to do. + // + // The keys are NOT moved retroactively, and that is a judgement about + // timing rather than a softening of the rule. A key bump invalidates + // entries written BEFORE the change; by 2026-10-28 the cache had already + // turned over on its own, verified rather than assumed: production + // serves 106 providers on hyperliquid-frontends including Perpex, and + // /products/perpex answers 200 carrying its hyperliquid-frontends + // appearance. Bumping now would flush healthy entries and surface + // nothing new. Had this been caught the same day, the keys would have + // had to move, and for the next change they still do. expect( hash, [ @@ -80,6 +90,6 @@ describe("the provider set behind the cache keys", () => { "Then set EXPECTED to the hash in this failure, and add the provider", `to PRODUCT_NAMES if it needs a display name. Providers now: ${count}.`, ].join("\n"), - ).toBe("f355324b6640e7e4"); + ).toBe("b767efc1bf40bf99"); }); });