mirrors/codex
1
0
Fork 0
mirror of https://github.com/openai/codex.git synced 2026-04-30 11:21:34 +03:00
Code Issues Packages Projects Releases Wiki Activity

exec-server: add in-process client mode

Browse Source 
Co-authored-by: Codex <noreply@openai.com>
This commit is contained in:
starr-openai
2026-03-17 20:10:22 +00:00
parent 17060efdbc
commit 4abc243801
9 changed files with 1023 additions and 347 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
codex-rs/exec-server/DESIGN.md
View File

@@ -70,7 +70,7 @@ Start simple and explicit:
Server notifications:
- `process/outputDelta`
- `process/output`
- `process/exited`
- optionally `process/started`
- optionally `process/failed`
@@ -132,7 +132,7 @@ growth.
Suggested behavior:
- stream every output chunk live via `process/outputDelta`
- stream every output chunk live via `process/output`
- retain capped output per process in memory
- keep stdout and stderr separately for pipe-backed processes
- for PTY-backed processes, treat retained output as a single terminal stream
@@ -234,7 +234,7 @@ If unified exec is later backed by exec-server, the `core` client wrapper should
keep owning the translation into the existing event model:
- `process/start` success -> `ExecCommandBegin`
- `process/outputDelta` -> `ExecCommandOutputDelta`
- `process/output` -> `ExecCommandOutputDelta`
- local `process/write` call -> `TerminalInteraction`
- `process/exited` plus retained transcript -> `ExecCommandEnd`