Commit Graph
360 Commits
Author SHA1 Message Date
pku-xht fc19a0a3fa fix(subprocess): clarify provider failures and scope polling 2026-08-31 19:26:08 +08:00
pku-xht 6d49ac2ef6 fix(subprocess): align runner bindings with current types 2026-08-31 15:37:14 +08:00
pku-xht fb00331cfa Merge master into subprocess native containment 2026-08-31 15:24:49 +08:00
pku-xht 5e3eaea742 refactor(subprocess): keep runner error bindings one-shot 2026-08-31 15:22:12 +08:00
Turtle b8e3b32fcf Merge master into codex/omit-unneeded-invariants 2026-08-31 14:38:55 +08:00
pku-xht 8bb19f8f7f fix(subprocess): close containment review findings 2026-08-31 14:36:13 +08:00
Yichen Jiang 44a9687bbe Merge remote-tracking branch 'origin/master' into worktree/deepseek-harness-proxy-config-2f5b4a
# Conflicts:
#	apps/cli/package.json
#	docs/module-graph.i18n.yaml
#	docs/module-graph.md
#	docs/module-graph.zh.md
#	packages/e2b/e2b/package.json
#	packages/llm/llm-deepseek/package.json
#	packages/llm/llm-pi-ai/package.json
#	packages/test-support/session-snapshot/package.json
#	packages/workflow/workflow-worker-thread/package.json
#	pnpm-lock.yaml
2026-08-31 12:58:59 +08:00
imccyu 3f1b46a5db release(dsh): 0.1.2-alpha.2 2026-08-30 21:19:29 +08:00
pku-xht 93c11689e7 fix(subprocess): address containment review findings 2026-08-29 15:44:13 +08:00
Yichen Jiang cfc9b3bdef fix(net): route by the policy, and give a child the routing its parent has
Second review pass on the outbound proxy work.

The installed dispatcher was undici's EnvHttpProxyAgent, which reuses the HTTP
proxy for `https:` whenever no HTTPS proxy is present. That is exactly the state
this package resolves after refusing a SOCKS or malformed URL the user named for
`https:`, so the scheme the diagnostic reported as direct was tunnelled anyway.
The dispatcher is now an Agent whose per-origin factory calls `proxyForUrl`, so
routing and `proxyForUrl` cannot disagree by parsing the same list twice.

`childProxyEnv` returned only the names the user exported, which left a child
Node direct whenever the proxy came from `ALL_PROXY` or from cordis.yml — Node's
`NODE_USE_ENV_PROXY` reads neither — and stripped the merged loopback bypass so
the child sent its own localhost traffic to the proxy. A scheme the user named
in either casing still reaches the child exactly as written; one they named in
neither now carries the resolved value, and the bypass list is always the merged
one.

A nested install (the plugin mounted over the launcher's policy) recorded the
outer policy's published values as the user's, then cleared the record on
disposal, so every later child inherited the normalization instead. The record
now belongs to the outermost install and is restored, not dropped.

`web_fetch` read the active policy twice — once to skip address pinning, again
inside the transport — so a disposal landing between the two reads produced an
unpinned direct connection to a host nothing validated. One snapshot now decides
both.

Also: the node:http proxy test asserted a route the engines range does not always
have, and the gate could not see undici bound through `await import('undici')`,
the form this repository actually uses.
2026-08-29 13:17:12 +08:00
Yichen Jiang e6dbf85f6c fix(net): address review — containment, opt-out, and syntax-aware discovery
The workflow worker no longer receives proxy configuration: it executes the
model-authored script body, and a proxy URL may carry credentials. A child
process now inherits the values the user exported rather than this process's
normalization, so a SOCKS proxy set for curl survives and no HTTPS_PROXY is
invented. `mode: 'off'` installs a direct dispatcher instead of recording a
policy the global dispatcher ignores, and the environment snapshot is taken
before any write so Windows restores the user's values.

E2B picks its proxy from the control-plane URL the SDK will really call, the
OTLP agent honors `exporter.keepAlive`, and a scheme whose own value was
refused stays direct instead of borrowing another scheme's proxy.

verify-no-bare-dispatcher parses the TypeScript AST as scripts/AGENTS.md
requires; it immediately found the `{ dispatcher }` shorthand the regex missed.
2026-08-29 13:17:12 +08:00
Yichen Jiang ec82e3e3ee test(net): assert the child and worker proxy seam by Node version
NODE_USE_ENV_PROXY reaches Node 24.0+ and 22.21+, while engines admits 22.19.
Assert the direct connection on an older runtime instead of only the proxied
one, so the seam is executable rather than prose.
2026-08-29 13:17:11 +08:00
Yichen Jiang 545e2ad914 feat(net): route every outbound request through the configured proxy
Node's built-in fetch ignores HTTP_PROXY, so every harness request connected
directly regardless of what the user exported. Resolve one policy from the
launch environment and install it as undici's global dispatcher, then wire the
four surfaces a global dispatcher cannot reach: web_fetch's pinned transport,
the OTLP exporter's node:http agent, the E2B SDK's own proxy option, and the
environment a child process or worker thread is given.

Each outbound call site carries an egress test that drives its real code path
through a fake proxy; that measurement is what found the OTLP and E2B gaps.
2026-08-29 13:17:11 +08:00
pku-xht aaa5117ebd fix(subprocess): preserve literal spawn error stacks 2026-08-29 13:08:43 +08:00
pku-xht c0e7b98169 test(subprocess): preserve optional spawn error keys 2026-08-29 13:02:36 +08:00
pku-xht 13f3b2c172 fix(subprocess): match native spawn error semantics 2026-08-29 12:13:30 +08:00
pku-xht 60dd357a02 fix(subprocess): preserve native spawn error shape 2026-08-29 11:36:27 +08:00
pku-xht 1760649ba7 refactor(subprocess): derive native owner state 2026-08-29 11:28:33 +08:00
pku-xht a416b13eb5 fix(subprocess): preserve Windows search miss errors 2026-08-29 10:28:44 +08:00
pku-xht 5aab793045 fix(subprocess): preserve clean Windows range settlement 2026-08-29 09:08:34 +08:00
pku-xht 6e7b152de7 fix(subprocess): preserve native runner contracts 2026-08-29 07:57:59 +08:00
pku-xht 23af4410f6 fix(subprocess): resolve source runner loader absolutely 2026-08-29 06:14:15 +08:00
pku-xht a825b8d042 fix(subprocess): close native containment contract gaps 2026-08-29 05:56:29 +08:00
pku-xht 40279aea17 fix(subprocess): preserve native runner stdio 2026-08-29 00:03:12 +08:00
pku-xht 635c35991c test(subprocess): type Linux exec mock 2026-08-28 22:38:18 +08:00
pku-xht a0bc2ea803 fix(subprocess): preserve native runner launch semantics 2026-08-28 22:34:09 +08:00
pku-xht aca5c0cb2b fix(subprocess): release Windows runner stdio owners 2026-08-28 21:29:09 +08:00
pku-xht 32b2b24099 fix(subprocess): close runner stdio handles 2026-08-28 20:30:11 +08:00
pku-xht 8ef28f76c0 fix(subprocess): preserve Windows runner spawn semantics 2026-08-28 19:55:16 +08:00
pku-xht 9aecbad9dd fix(subprocess): keep runner worker-loadable 2026-08-28 19:16:34 +08:00
pku-xht b65575442d Merge origin/master into codex/subprocess-native-containment 2026-08-28 18:34:22 +08:00
pku-xht a95f0b368f fix(subprocess): contain escaped descendants with native owners 2026-08-28 18:15:49 +08:00
Turtle 15f2997bcb cleanup: omit unneeded invariant companions 2026-08-28 13:12:52 +08:00
imccyu 6c705be1ce release(dsh): 0.1.2-alpha.1 2026-08-28 00:50:12 +08:00
Yichen Jiang 9757224349 fix(subprocess): fence each signal against current process state
Review found the shared observation defeated the very fence it fed:
it carries the original PID-to-start-time pairing forward, so a
recycled PID still matches it and takes a signal meant for the process
that exited. Capturing it outside the per-member try also let one
failed read abort a whole teardown round, breaking the synchronous
host-exit contract, and an empty round paid a read for no members.

signalProcess now reads ProcessInspector.isAlive immediately before
delivering, from the narrowest per-identity source each platform
offers; signalMembers and waitForMembers return before capturing when
a round has no members. snapshot() keeps serving the readiness poll,
whose per-poll table read stays at one.

Windows enumerates Toolhelp32 lazily on the first tree question, so a
snapshot asked only for liveness — the 25 ms teardown poll — performs
no table walk at all.
2026-08-27 15:19:43 +08:00
Yichen Jiang 32ddfcd89c fix(subprocess): read the process table once per terminal poll
MacProcessInspector answered the descendant tree and every member's
liveness with its own `/bin/ps` fork, so one readiness poll cost N+1
full table reads for N tracked descendants. With execFileSync on that
path and a 50 ms poll interval, any command spawning two or more
children saturated the host event loop until it exited.

ProcessInspector.snapshot() now returns one ProcessSnapshot that
answers tree, session, and alive from a single observation, and
signalProcess takes the caller's observation so its PID-reuse fence
does not re-read the table per member.
2026-08-27 13:06:12 +08:00
pku-xht c302bedc3a test(subprocess): defer fake runner disconnect 2026-08-27 02:37:00 +08:00
pku-xht ef0815db7f chore(subprocess): document disconnect recheck 2026-08-27 01:33:06 +08:00
pku-xht 1250b41054 docs(subprocess): clarify Windows runner disconnect 2026-08-27 00:57:21 +08:00
pku-xht 81257defd2 fix(subprocess): preserve clean Windows runner close 2026-08-27 00:41:59 +08:00
pku-xht 07408ff0f1 fix(subprocess): preserve failed teardown state 2026-08-26 23:43:15 +08:00
pku-xht 3caa9e4f5a fix(subprocess): retain failed cleanup ownership 2026-08-26 22:16:20 +08:00
pku-xht 166004a6a8 fix(subprocess): preserve Windows startup errors 2026-08-26 20:01:10 +08:00
pku-xht 6a34a0dd4b fix(subprocess): isolate Linux native target group 2026-08-26 19:00:40 +08:00
pku-xht 20f04373c8 Merge commit 'bd88d2ee9f08b3c3aba8c06df527a91d8b76b51d' into codex/subprocess-native-containment
# Conflicts:
#	packages/e2b/subprocess-e2b/README.i18n.yaml
#	packages/e2b/subprocess-e2b/README.md
#	packages/e2b/subprocess-e2b/README.zh.md
#	packages/subprocess/README.i18n.yaml
#	packages/subprocess/README.md
#	packages/subprocess/README.zh.md
#	packages/subprocess/subprocess-local/README.i18n.yaml
#	packages/subprocess/subprocess-local/README.md
#	packages/subprocess/subprocess-local/README.zh.md
#	packages/subprocess/subprocess/README.i18n.yaml
#	packages/subprocess/subprocess/README.md
#	packages/subprocess/subprocess/README.zh.md
#	packages/subprocess/win32-process/README.i18n.yaml
#	packages/subprocess/win32-process/README.md
#	packages/subprocess/win32-process/README.zh.md
2026-08-26 15:25:12 +08:00
pku-xht cf6f25d283 fix(subprocess): keep native runner private 2026-08-26 14:10:08 +08:00
Magolor 0b5eba0c8d docs: rebuild the documentation skill and standards (#2983) 2026-08-25 23:47:20 +08:00
pku-xht 678cfdfdb4 Merge commit '0c177e17a23692d018a79bae59ac9a2db6eba59c' into codex/subprocess-native-containment
# Conflicts:
#	.agents/notes/implemented/architecture/2026-07-10-single-file-executable-sdk-runtime-distribution.i18n.yaml
#	.agents/notes/implemented/architecture/2026-07-10-single-file-executable-sdk-runtime-distribution.md
#	.agents/notes/implemented/architecture/2026-07-10-single-file-executable-sdk-runtime-distribution.zh.md
#	examples/acp-agent/tests/snapshots/cordis-inspect-jsdoc/input.json
#	examples/acp-agent/tests/snapshots/cordis-inspect-jsdoc/session.jsonl
#	packages/bundle/acp-app/package.json
#	packages/examples/jsonrpc-demo/src/packaged-bin.ts
#	packages/subagent/subagent-acp/src/run.ts
#	packages/subagent/subagent-acp/tests/subagent-acp.spec.ts
#	packages/subprocess/subprocess-local/README.i18n.yaml
#	packages/subprocess/subprocess-local/README.md
#	packages/subprocess/subprocess-local/README.zh.md
#	pnpm-lock.yaml
#	scripts/smoke-python-runtime.py
2026-08-25 23:22:25 +08:00
pku-xht ff413cfe5e fix(subprocess): align async pid evidence 2026-08-25 22:24:20 +08:00
pku-xht 26969588b7 fix(subprocess): address native containment review findings 2026-08-25 21:51:22 +08:00