Skip to content

New sessions spawn under holders - #102

Merged
karngyan merged 1 commit into
mainfrom
holder-3-spawn
Aug 19, 2026
Merged

New sessions spawn under holders#102
karngyan merged 1 commit into
mainfrom
holder-3-spawn

Conversation

@karngyan

Copy link
Copy Markdown
Contributor

Third in the holder-sessions stack, on top of #101.

flue serve points the registry at the holder executable and <config>/holders, so every Spawn and Revive now runs its session out-of-process. FLUE_NO_HOLDER=1 is the escape hatch back to in-process sessions, and tests keep exercising the engine directly. Each holder directory carries the daemon-owned session.json identity record (id, display cmd, group, ephemeral, createdAt); the keep affordance's ephemeral flip is persisted there so a later reattach honors it.

Smoke-tested against the real binary: spawn over the ws wire, then kill -9 the daemon — the holder and the session's child keep running. (Reattach on daemon start is the next PR; until then a restarted daemon does not yet pick live holders back up.)

🤖 Generated with Claude Code

Base automatically changed from holder-2-holder to main August 19, 2026 08:58
Spawn and Revive run each session under its own holder once the
registry is pointed at the holder executable and root, which flue
serve now does; FLUE_NO_HOLDER=1 keeps the in-process path. The
identity record written beside each socket is the daemon-owned half
a reattach needs, and the keep affordance's ephemeral flip lands
there too.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
@karngyan
karngyan merged commit 8fddebb into main Aug 19, 2026
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant