fix(watch): apply Transition payload in place — stop full-list redraws#31
Merged
Conversation
User-reported: STATE / LAST PROMPT columns appear to "update for
every row" on every tick, even though only one agent had a real
transition. Asked: can we skip the update for rows that didn't have
an event?
Yes — the v0.5.0 subscribe stream was being treated as a wake
signal only. Every Transition triggered a fresh `client.snapshot()`
and `apply_outcome` replaced the whole agent list. Even when only
one row's data actually changed, every row was rewritten with
"current values" — visually indistinguishable from constant churn.
Restructure:
- `RefreshOutcome` becomes an enum:
- `Full(FullRefresh)` — periodic fallback (5 s) and priming wake;
replaces the whole list (existing behavior, isolated path).
- `SingleAgent(Agent)` — push delivery; carries just the
transitioned agent's post-transition payload.
- `refresh_task`'s subscribe arm ships `Transition.agent` directly
via `SingleAgent` instead of falling through to `fetch().await`.
- `apply_outcome` matches on the variant. The new
`apply_single_agent` finds the row by `(kind, session_id)` and
replaces only that one; appends if unknown. No re-sort on push so
the table doesn't visually "jump" — sort happens in the periodic
Full tick.
Performance: push deliveries no longer pay the round-trip cost of a
snapshot RPC + tmux pane re-fetch. Only the matching row's bytes
change. The fallback Full tick still keeps everything else fresh.
Tests: 2 new — `single_agent_outcome_only_updates_matching_row` and
`single_agent_outcome_appends_unknown_row`. Existing tests that
constructed `RefreshOutcome` directly were updated to use
`RefreshOutcome::Full(FullRefresh { … })`. Workspace 425 → 427 pass.
clippy clean (with one targeted `large_enum_variant` allow on the
enum — boxing the Full variant would add an allocation per tick for
no real-world benefit). fmt clean.
No version bump.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
User reported: STATE / LAST PROMPT in `muxa watch` appear to update for every row on every tick even when only one agent had an event. Asked if we could skip updates for rows without their own transition.
Yes — v0.5.0's subscribe was only used as a wake signal: each Transition triggered a fresh `client.snapshot()` and `apply_outcome` replaced the entire agent list. The single row that changed got the correct new value, but every other row was also rewritten with "current values" — visually indistinguishable from constant churn.
Fix
Push deliveries no longer pay the snapshot RPC + tmux pane re-fetch round-trip. Other rows keep their prior bytes.
Test plan
🤖 Generated with Claude Code