mirror of
https://github.com/badlogic/pi-mono.git
synced 2026-11-05 23:37:25 +00:00
5f312dac4c
Chord's shared watch: listeners take a tagged call, { kind: "value", value, ops, retired } or
{ kind: "failed", error }; a recoverable failure keeps the watch open and the next value arrives as
[["r", value]]; WatchEnd is tagged by kind, and an aborted context ends a watch "aborted".
Durable: observers take the watch's shapes (advance(call, ctx), end(end)); onReport(report) with
report.kind and report.error, session_failed is storage_failed; subscribeCommits is observeCommits;
SessionEnd and ReportFrom are tagged by kind; FileChange is tagged and file watchers stop with
stop(ctx); CompactionFailure and CompactionStatus name their trigger (pi.live version 2 with a
migrate); the task_outdated report uses the scheduler's reasons; TaskInspection's ready is pending;
DocumentCommitChange is tagged. Stored formats are unchanged.
96 lines
4.1 KiB
TypeScript
96 lines
4.1 KiB
TypeScript
/**
|
|
* Durable code mode overhead. One model-issued call makes N calls of a trivial `echo` tool, either sequentially or all at
|
|
* once, in two ways: a code mode script (`tools.echo()`, journaled and replayable) and a plain tool calling
|
|
* `api.executeTool()` the same way (nested calls, no journal). Reports wall time and commits per call on SQLite.
|
|
*
|
|
* node --conditions=source --experimental-strip-types test/codemode-bench.ts [N,...]
|
|
*/
|
|
import { mkdtemp, rm } from "node:fs/promises";
|
|
import { tmpdir } from "node:os";
|
|
import { join } from "node:path";
|
|
import { BACKGROUND_CONTEXT as context } from "@earendil-works/chord/context";
|
|
import { createModels } from "@earendil-works/pi-ai/models";
|
|
import { fauxAssistantMessage, fauxProvider, fauxText, fauxToolCall } from "@earendil-works/pi-ai/providers/faux";
|
|
import { CodemodeSandbox } from "@earendil-works/pi-codemode";
|
|
import { Type } from "typebox";
|
|
import { createCodemodeTool } from "../src/codemode/codemode.ts";
|
|
import { createRegistry, defineExtension, defineTool, Harness } from "../src/index.ts";
|
|
import { openNodeSqliteStorage } from "../src/storage/sqlite/node.ts";
|
|
|
|
type Shape = "sequential" | "parallel";
|
|
|
|
const echo = defineTool({
|
|
name: "echo",
|
|
description: "echo",
|
|
parameters: Type.Object({ i: Type.Number() }),
|
|
execute: async (args) => ({ output: [{ type: "text", text: String(args.i) }] }),
|
|
});
|
|
|
|
const plain = defineTool({
|
|
name: "plain",
|
|
description: "plain",
|
|
parameters: Type.Object({ n: Type.Number(), shape: Type.String() }),
|
|
replay: "safe",
|
|
execute: async (args, api, callContext) => {
|
|
if (args.shape === "sequential") {
|
|
for (let i = 0; i < args.n; i++) await api.executeTool("echo", { i }, callContext);
|
|
} else {
|
|
await Promise.all(Array.from({ length: args.n }, (_, i) => api.executeTool("echo", { i }, callContext)));
|
|
}
|
|
return { output: [{ type: "text", text: "done" }] };
|
|
},
|
|
});
|
|
|
|
async function run(mode: "codemode" | "plain", shape: Shape, n: number): Promise<{ ms: number; commits: number }> {
|
|
const directory = await mkdtemp(join(tmpdir(), "pi-codemode-bench-"));
|
|
try {
|
|
const faux = fauxProvider();
|
|
const models = createModels();
|
|
models.setProvider(faux.provider);
|
|
const sandbox = new CodemodeSandbox({ memoryLimitBytes: 256 * 1024 * 1024 });
|
|
const registry = createRegistry();
|
|
registry.install(defineExtension({ name: "bench", tools: [echo, plain, createCodemodeTool({ sandbox })] }));
|
|
const code =
|
|
shape === "sequential"
|
|
? `for (let i = 0; i < ${n}; i++) await tools.echo({ i }); return "done";`
|
|
: `await Promise.all(Array.from({ length: ${n} }, (_, i) => tools.echo({ i }))); return "done";`;
|
|
const call =
|
|
mode === "codemode"
|
|
? fauxToolCall("codemode", { code }, { id: "c1" })
|
|
: fauxToolCall("plain", { n, shape }, { id: "c1" });
|
|
faux.setResponses([
|
|
fauxAssistantMessage([call], { stopReason: "toolUse" }),
|
|
fauxAssistantMessage([fauxText("ok")]),
|
|
]);
|
|
const storage = await openNodeSqliteStorage(join(directory, "session.sqlite"));
|
|
const harness = await Harness.open(storage, { models, registry, settings: {} }, context);
|
|
const root = await harness.root(context, { agent: { model: { provider: "faux", modelId: "faux-1" } } });
|
|
let commits = 0;
|
|
harness.observeCommits({
|
|
advance: () => {
|
|
commits++;
|
|
},
|
|
end: () => {},
|
|
});
|
|
const started = performance.now();
|
|
await (await root.submit({ type: "input", content: "go" }, context)).wait(context);
|
|
const ms = performance.now() - started;
|
|
await harness.close(context);
|
|
await sandbox.close();
|
|
return { ms, commits };
|
|
} finally {
|
|
await rm(directory, { recursive: true, force: true });
|
|
}
|
|
}
|
|
|
|
const sizes = (process.argv[2] ?? "100,1000").split(",").map(Number);
|
|
for (const shape of ["sequential", "parallel"] as const) {
|
|
for (const n of sizes) {
|
|
const plainRun = await run("plain", shape, n);
|
|
const codemodeRun = await run("codemode", shape, n);
|
|
const per = (value: number) => (value / n).toFixed(2);
|
|
console.log(
|
|
`${shape} n=${n}: plain ${plainRun.ms.toFixed(0)} ms (${per(plainRun.ms)} ms, ${per(plainRun.commits)} commits/call), codemode ${codemodeRun.ms.toFixed(0)} ms (${per(codemodeRun.ms)} ms, ${per(codemodeRun.commits)} commits/call)`,
|
|
);
|
|
}
|
|
}
|