Keep OSC 7501 program-status records per session and list them #128

Merged

natsukium merged 3 commits from daemon/osc7501-records into main 2026-10-12 23:04:40 +09:00

Conversation 0 Commits 3 Files changed 50 +2084 -52

natsukium commented 2026-10-12 21:57:32 +09:00

Owner

Copy link

Fourth and last step of #67. The daemon now keeps OSC 7501 program-status records per session and exposes them on every SessionInfo as program_status, so felis sessions list --format json, felis sessions info, the bridge and any roster reader see them. felis still draws nothing; support is experimental (spec revision 0.4).

• OSC 7501 ; ? is answered with the bare reply, attached or not.
• Records follow the spec: replace whole, clear removes the subtree, at most 256 per session with least-recently-updated eviction. A primary prompt start and PTY end (after the child's last bytes are parsed) drop working/blocked; RIS drops all. Records do not cross an in-place upgrade.
• SessionInfo.program_status joins protocol minor 1 under the new ungated-field rule (#127): declared in ADDED_FIELDS, sent to every peer, a minor-0 peer skips it.
• A roster reply that would exceed the frame limit is refused with an error naming its size and felis sessions info <id>, instead of a frame that drops the connection. Reaching it needs hundreds of sessions each full of maximal records.
• Two existing paths fixed on the way: a failed PTY reply no longer discards the effects queued behind it, and roster-bearing Ops replies are charged by content against the subscriber backlog instead of a flat 64 bytes.
• Human sessions info output spells bidi and control characters in producer text as <U+XXXX>; IPC and JSON carry the decoded text.

Docs: new reference and explanation pages for program status, support-matrix row, ipc.md ledger and field, cli.md, the product skill, CHANGELOG.

Checks: just check green; cargo-mutants on the store and the oversize path, 0 missed; pi design review (3 rounds, direction then simplified per review) and code review PASS.

Closes #67.

Fourth and last step of #67. The daemon now keeps OSC 7501 program-status records per session and exposes them on every `SessionInfo` as `program_status`, so `felis sessions list --format json`, `felis sessions info`, the bridge and any roster reader see them. felis still draws nothing; support is experimental (spec revision 0.4). - `OSC 7501 ; ?` is answered with the bare reply, attached or not. - Records follow the spec: replace whole, `clear` removes the subtree, at most 256 per session with least-recently-updated eviction. A primary prompt start and PTY end (after the child's last bytes are parsed) drop `working`/`blocked`; RIS drops all. Records do not cross an in-place upgrade. - `SessionInfo.program_status` joins protocol minor 1 under the new ungated-field rule (#127): declared in `ADDED_FIELDS`, sent to every peer, a minor-0 peer skips it. - A roster reply that would exceed the frame limit is refused with an error naming its size and `felis sessions info <id>`, instead of a frame that drops the connection. Reaching it needs hundreds of sessions each full of maximal records. - Two existing paths fixed on the way: a failed PTY reply no longer discards the effects queued behind it, and roster-bearing Ops replies are charged by content against the subscriber backlog instead of a flat 64 bytes. - Human `sessions info` output spells bidi and control characters in producer text as `<U+XXXX>`; IPC and JSON carry the decoded text. Docs: new reference and explanation pages for program status, support-matrix row, ipc.md ledger and field, cli.md, the product skill, CHANGELOG. Checks: `just check` green; cargo-mutants on the store and the oversize path, 0 missed; pi design review (3 rounds, direction then simplified per review) and code review PASS. Closes #67.

natsukium added 2 commits 2026-10-12 21:57:32 +09:00

grid: answer the OSC 7501 detection query ... 7f8c95b33b

A program learns that the terminal keeps its status only from the
reply to `OSC 7501 ; ?`, and the spec recommends DA1 right after the
query so silence costs nothing; with the records now kept and listed
there is support to announce. The reply is bare: revision 0.4 reads a
reply without `states=` / `kinds=` as exactly the base vocabulary,
which is what the decoder admits. It always ends in ST, as the OSC 99
`p=?` reply does, and a cancelled query is not answered.

Assisted-by: Claude Code

protocol, grid, daemon, cli: keep OSC 7501 records per session and list them ... 3a28a68371

A TUI agent waiting on its user has no signal felis can read: it sets
no OSC 133 marks, so `sessions send --wait` never fires. OSC 7501 is the
producer saying so itself. The grid already queues reports, prompt
starts and RIS on the ordered effect queue; this applies them in the
session task and puts the records on every SessionInfo, so `sessions
list --format json`, `sessions info`, the bridge and any roster reader
see them. Support is experimental and draws nothing.

The store holds the spec's 256 records per session, so a producer that
follows the spec is never cut short by felis. The cost is that one
`Listed` frame carries every session's records and can, in principle,
outgrow the frame limit; no real roster comes near it. Rather than
writing a frame the peer must drop, which ends the connection with no
explanation, the daemon now refuses such a reply with the roster's size
and points at `felis sessions info <id>`, which reads one session.

The field joins protocol minor 1: no released peer speaks minor 1, so
no peer can lack the field while having the rest of that minor. It is
sent to every peer, and a minor-0 one skips it as an unknown field.

Two existing paths would lose a record otherwise:

- `replay_effects` returned on the first failed PTY write after taking
  the whole queue, so a `done` written right before exit, behind a
  query reply the dead PTY refused, vanished; the replay now records
  the first write error and keeps applying state effects.
- An `Ops` reply was charged a flat 64 bytes against the subscriber
  backlog, so a peer listing repeatedly while reading slowly could
  queue rosters far past the cap; roster-bearing replies are now
  charged by their contents.

PTY EOF expires transient records after the drive that parses the
child's last output, so the final report survives the expiry. Records
are dropped on an in-place upgrade. The human `sessions info` line
spells direction controls in producer text as `<U+XXXX>`; IPC and JSON
carry the decoded value.

Assisted-by: Claude Code

natsukium added 1 commit 2026-10-12 22:32:26 +09:00

docs, skills: lead the program-status docs with what a user does ... ba56043c2f

The program-status docs described the wire and the IPC shape but gave a
user no route to the task they come with: finding the session whose
program waits on them. Nothing linked the new pages from the doc map,
the reference opened with IPC terms, and the only polling recipe, in
the product skill, waited on the root record alone, so a program that
exits without a final report or reports under a sub-id held the loop to
its deadline, and the loop said nothing when that happened.

A how-to now covers checking that a program reports, listing blocked
sessions, waiting for one with an exit status per outcome, reporting
from one's own program (with detection), and clearing a stale record.
The reference states the rejection rules as the decoder applies them
(an over-long app discards the report; a value with a byte outside the
value charset is skipped), says that prompt expiry needs shell marks,
and that nothing announces a change. The skill drops producer names
nothing in the repository confirms.

Assisted-by: Claude Code

natsukium merged commit b7f9361596 into main 2026-10-12 23:04:40 +09:00

natsukium deleted branch daemon/osc7501-records 2026-10-12 23:04:40 +09:00

natsukium referenced this pull request from a commit 2026-10-12 23:04:42 +09:00

Merge pull request 'Keep OSC 7501 program-status records per session and list them' (#128) from daemon/osc7501-records into main

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels   

Area/ci

CI workflows (.forgejo/workflows/)

  

Area/cli

felis-cli crate, the felis front-door binary

  

Area/client

felis-client crate, the winit GUI frontend

  

Area/client-core

felis-client-core crate, shadow screen, config, connect_or_spawn_daemon

  

Area/daemon

felis-daemon crate, session pool, IPC server

  

Area/deps

Dependency bumps

  

Area/docs

Documentation tree when no narrower area fits

  

Area/grid

felis-grid crate, cells, scrollback, image store

  

Area/manual

User-facing docs: tutorials, how-to, user-facing reference pages

  

Area/nix

Nix flake, home-manager module, package definitions

  

Area/protocol

felis-protocol crate, felis.proto, frame layout, version constants

  

Area/pty

felis-pty crate, PTY backends

  

Area/render

felis-render-wgpu crate, GPU pipeline, atlas

  

Area/shaping

felis-shaping crate, swash shaper, font fallback

  

Area/skills

Agent skills (skills/felis, .agents/skills/)

  

Area/transport

felis-transport crate, framing, socket/pipe, SSH carrier

  

Area/vt

felis-vt crate, VT parser, dispatch

  

Compat/Breaking

Breaking change that won't be backward compatible

  

Kind/Bug

Something is not working

  

Kind/Documentation

Documentation changes

  

Kind/Enhancement

Improve existing functionality

  

Kind/Feature

New functionality

  

Kind/Performance

Throughput, latency, or memory cost

  

Kind/Security

This is security issue

  

Kind/Testing

Issue or pull request related to testing

  

perf-regression-accepted

Passes the instruction-count bench gate for an intended cost; state why in a PR comment

  

Reviewed

Confirmed

Issue has been confirmed

  

Reviewed

Duplicate

This issue or pull request already exists

  

Reviewed

Invalid

Invalid issue

  

Reviewed

Won't Fix

This issue won't be fixed

  

Status

Blocked

Something is blocking this issue or pull request

  

Status

Need More Info

Feedback is required to reproduce issue or to continue work

  

Status

Upstream

Waiting on a change in an upstream project

No labels

Area/ci

Area/cli

Area/client

Area/client-core

Area/daemon

Area/deps

Area/docs

Area/grid

Area/manual

Area/nix

Area/protocol

Area/pty

Area/render

Area/shaping

Area/skills

Area/transport

Area/vt

Compat/Breaking

Kind/Bug

Kind/Documentation

Kind/Enhancement

Kind/Feature

Kind/Performance

Kind/Security

Kind/Testing

perf-regression-accepted

Reviewed

Confirmed

Reviewed

Duplicate

Reviewed

Invalid

Reviewed

Won't Fix

Status

Blocked

Status

Need More Info

Status

Upstream

Milestone

Clear milestone

No items

No milestone

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

felis-terminal/felis!128

No description provided.