hermes-agent

mirrors/hermes-agent

Fork 0

mirror of https://github.com/NousResearch/hermes-agent.git synced 2026-07-28 18:19:28 +00:00

Commit graph

Author	SHA1	Message	Date
Teknium	1927b60775	fix(tui): extend deferred context-engine finalize to the compute-host compress routes The salvaged deferred-finalize wiring (#65670) covered cli.py, the gateway slash command, and the three in-process tui_gateway/server.py compress sites — but the dashboard compute-host isolated session.compress / slash.compress routes run the compress mirror inside the host child, where a control-handler exception after compress_context() queued the boundary notification would leak it: never fired on success paths, or worse, fired by a LATER compress against a boundary the host had rejected. - tui_gateway/compute_host.py: _handle_control discards any pending context-engine compression notification (committed=False) when a session.compress / slash.compress control frame errors. finalize is exactly-once, so this is a no-op when the mirror already emitted or discarded it. - tui_gateway/server.py: _mirror_slash_side_effects normalizes the /compact alias onto the compress branch so isolated-session compacts actually compress and hit the same finalize/discard wiring instead of silently no-oping. - tests/tui_gateway/test_compute_host_phase1.py: compute-host route tests — commit-then-notify ordering, discard on host commit failure, /compact alias routing.	2026-07-22 06:58:16 -07:00
x7peeps	01c02c6f8f	fix(tui): recover from spurious stdin EOF caused by child O_NONBLOCK flip Fix #67639 根因分析: 子进程继承 fd 0 (stdin) 后设置 O_NONBLOCK 标志时，该标志作用于共享的 open file description 而非单个文件描述符。这导致 gateway 的下一次 read() 返回 EAGAIN，CPython 的缓冲层将其转换为 b''，表现为 EOF，gateway 因此意外退出。这不是真正的 TUI 关闭管道，而是子进程修改了共享文件状态。修复涉及三个 Gap: Gap 1 — check_subprocess_stdin.py 扫描范围不足: - 原正则仅匹配 subprocess.run/Popen，扩展到 call/check_output/check_call/ os.system/asyncio.create_subprocess_exec/shell - 新增扫描 ~/.hermes/plugins/ 和 ./.hermes/plugins/ 用户插件目录 (hermes_cli/plugins.py:10-12 定义的插件加载路径) Gap 2 — Gateway 无自愈能力: - entry.py 和 slash_worker.py 的 stdin 循环改为 while True + readline() 模式 - 当 readline() 返回空字符串时，检查 O_NONBLOCK 标志判断是否为虚假 EOF - 虚假 EOF → 恢复 blocking 模式并继续; 真实 EOF → 正常退出 - 添加恢复频率限制 (10次/分钟)，防止无限循环 - 添加 _diagnose_stdin_state() 诊断函数，记录 O_NONBLOCK/SO_RCVTIMEO 状态 Gap 3 — 日志消息误导: - 原 "stdin EOF (TUI closed the command pipe)" 改为 "stdin EOF (peer closed)" 或 "stdin spurious EOF (subprocess O_NONBLOCK flip)"，附带诊断信息附带修复: compute_host.py 的 subprocess.check_output 缺少 stdin= 参数	2026-07-20 12:48:50 +05:30
brooklyn!	7d27a31ce7	feat(dashboard): isolate turns in compute host (#65895 ) Add the flag-gated compute-host supervisor, delta/control protocol, PPID orphan guard, inline fail-open path, synthetic GIL-heavy turn seam, and AC-4 certify harness. Verified on current origin/main: 343 focused tests pass; Ruff and diff checks pass; 360s AC-4 run with six heavy lanes passes at 6.11ms serving p99 with zero stalls and valid load. Co-authored-by: Kyzcreig <9063726+Kyzcreig@users.noreply.github.com>	2026-07-16 15:24:03 -04:00

Author

SHA1

Message

Date

Teknium

1927b60775

fix(tui): extend deferred context-engine finalize to the compute-host compress routes

The salvaged deferred-finalize wiring (#65670) covered cli.py, the gateway
slash command, and the three in-process tui_gateway/server.py compress
sites — but the dashboard compute-host isolated session.compress /
slash.compress routes run the compress mirror inside the host child, where
a control-handler exception after compress_context() queued the boundary
notification would leak it: never fired on success paths, or worse, fired
by a LATER compress against a boundary the host had rejected.

- tui_gateway/compute_host.py: _handle_control discards any pending
  context-engine compression notification (committed=False) when a
  session.compress / slash.compress control frame errors. finalize is
  exactly-once, so this is a no-op when the mirror already emitted or
  discarded it.
- tui_gateway/server.py: _mirror_slash_side_effects normalizes the
  /compact alias onto the compress branch so isolated-session compacts
  actually compress and hit the same finalize/discard wiring instead of
  silently no-oping.
- tests/tui_gateway/test_compute_host_phase1.py: compute-host route tests —
  commit-then-notify ordering, discard on host commit failure, /compact
  alias routing.

2026-07-22 06:58:16 -07:00

x7peeps

01c02c6f8f

fix(tui): recover from spurious stdin EOF caused by child O_NONBLOCK flip

Fix #67639

根因分析:
子进程继承 fd 0 (stdin) 后设置 O_NONBLOCK 标志时，该标志作用于共享的
open file description 而非单个文件描述符。这导致 gateway 的下一次 read()
返回 EAGAIN，CPython 的缓冲层将其转换为 b''，表现为 EOF，gateway 因此
意外退出。这不是真正的 TUI 关闭管道，而是子进程修改了共享文件状态。

修复涉及三个 Gap:

Gap 1 — check_subprocess_stdin.py 扫描范围不足:
- 原正则仅匹配 subprocess.run/Popen，扩展到 call/check_output/check_call/
  os.system/asyncio.create_subprocess_exec/shell
- 新增扫描 ~/.hermes/plugins/ 和 ./.hermes/plugins/ 用户插件目录
  (hermes_cli/plugins.py:10-12 定义的插件加载路径)

Gap 2 — Gateway 无自愈能力:
- entry.py 和 slash_worker.py 的 stdin 循环改为 while True + readline()
  模式
- 当 readline() 返回空字符串时，检查 O_NONBLOCK 标志判断是否为虚假 EOF
- 虚假 EOF → 恢复 blocking 模式并继续; 真实 EOF → 正常退出
- 添加恢复频率限制 (10次/分钟)，防止无限循环
- 添加 _diagnose_stdin_state() 诊断函数，记录 O_NONBLOCK/SO_RCVTIMEO 状态

Gap 3 — 日志消息误导:
- 原 "stdin EOF (TUI closed the command pipe)" 改为 "stdin EOF (peer closed)"
  或 "stdin spurious EOF (subprocess O_NONBLOCK flip)"，附带诊断信息

附带修复: compute_host.py 的 subprocess.check_output 缺少 stdin= 参数

2026-07-20 12:48:50 +05:30

brooklyn!

7d27a31ce7

feat(dashboard): isolate turns in compute host (#65895 )

Add the flag-gated compute-host supervisor, delta/control protocol, PPID orphan guard, inline fail-open path, synthetic GIL-heavy turn seam, and AC-4 certify harness.

Verified on current origin/main: 343 focused tests pass; Ruff and diff checks pass; 360s AC-4 run with six heavy lanes passes at 6.11ms serving p99 with zero stalls and valid load.

Co-authored-by: Kyzcreig <9063726+Kyzcreig@users.noreply.github.com>

2026-07-16 15:24:03 -04:00

3 commits