mirror of
https://github.com/diegosouzapw/OmniRoute.git
synced 2026-09-13 18:32:12 +03:00
65e81158ab
Landed with the design call resolved per the owner's pick — **option 1**: the synced store is now endpoint-agnostic (persistDiscoveredModels and managedModelImport no longer drop non-chat models at write time), and chat selectability moved to read time (auto-pool expansion in autoStrategy applies filterChatSelectableModels; the models-route projection already had its chatOnly filter). Your discovery test now passes end-to-end (3/3): /api/show capabilities persist per connection and image/embedding requests route through the advertising host. Reconciliation notes: conflicted areas merged onto the current tip (adobe discovery import, requestedModel preflight signature, resolvedProvider fast-path coexists with the synced-route override — explicit resolution wins); carried base-red drains (#10055 memoization, #11071 test variants) dropped as already-landed; the managed-model-import exclusion test was propagated to the new contract (image/video models persist; the read filter still hides them from chat pickers — pinned by a new assertion). Full battery: 205/206 focused (the one red is a confirmed periodic-timer timing flake on the loaded devbox — 20/20 isolated), autoCombo vitest 30/30, combo suites 46/46, gates + typecheck clean. Thank you @yourspraveen — the capability probe + routing design was right; it just needed the store contract opened up. Fixes #11087.
28 lines
1.6 KiB
TypeScript
28 lines
1.6 KiB
TypeScript
import { test } from "node:test";
|
|
import assert from "node:assert/strict";
|
|
import { requiresReasoningReplay } from "../../open-sse/services/reasoningCache.ts";
|
|
|
|
// Regression guard: the reasoning-cache write in chatCore.ts must be gated by the same
|
|
// predicate the read sites already use, so an installation whose traffic never touches a
|
|
// replay provider stops writing to the cache on every reasoning-bearing response.
|
|
test("requiresReasoningReplay recognizes every provider chatCore.ts's write guard must cover", () => {
|
|
// Providers explicitly in REASONING_REPLAY_PROVIDERS
|
|
assert.equal(requiresReasoningReplay({ provider: "deepseek", model: "deepseek-chat" }), true);
|
|
assert.equal(requiresReasoningReplay({ provider: "kimi-coding", model: "kimi-k2" }), true);
|
|
assert.equal(requiresReasoningReplay({ provider: "xiaomi-mimo", model: "mimo-v1" }), true);
|
|
|
|
// Providers NOT in the set, caught by the model-pattern fallback (the same fallback
|
|
// that covers claudeHelper.ts's non-Anthropic Claude-shape read site, site C in the
|
|
// design decision above)
|
|
assert.equal(
|
|
requiresReasoningReplay({ provider: "some-glm-host", model: "glm-4-thinking" }),
|
|
true
|
|
);
|
|
assert.equal(requiresReasoningReplay({ provider: "zai", model: "kimi-k2" }), true);
|
|
|
|
// A provider/model that should NOT trigger a write — proves the guard actually skips
|
|
// something (otherwise this write-guard change is a no-op)
|
|
assert.equal(requiresReasoningReplay({ provider: "openai", model: "gpt-5.1" }), false);
|
|
assert.equal(requiresReasoningReplay({ provider: "anthropic", model: "claude-opus-5" }), false);
|
|
});
|