Repository navigation
RAL-202: the roster shows live sessions first and folds stale ones away - #291
Merged
Merged
Conversation
The 7300 roster was full of hook-only leftovers (7 of the live roster's rows), shown in project order and mixed in with live sessions. - shims::stale: a row is stale when the daemon records it as bound to no connection, or its declared pid is 0, and the daemon holds no wait for it. The reason says which fact, with its label. - The viewer puts live sessions first, most recently active first. Stale rows fold under one "N stale sessions" line, closed until clicked (kept open across redraws by data-key). Each one says why it's stale, and none says "connected N ago". A live row that was never bound reads "on the roster since". A stale "busy" never shows under "Now". - "since unknown" becomes "last active N ago" where the roster has a time and "since: not reported by its roster" where it doesn't. It never reads "for" an age taken from the last tool (RAL-124 holds). - stalls, wake and brief drop stale rows before matching teams, and their reports count them as stale_sessions. - Unit tests of the rule, a headless Chrome test of the panel, and the RAL-116 and RAL-124 browser tests updated to the new wording. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The watchman's follow-up to part 2. A hook-only ghost carrying the designated name ended the digest's route early at a dead session, and a ghost sharing a live team's nickname made that team "ambiguous". - marshal::recipient matches a nickname only against live sessions: bound to a shim's connection, or holding a wait the daemon has now. When only a stale row holds the name, it says so. An exact session id is still taken as given. - send_at also watches the wake watchers, so a hook-only session that is waiting can still be reached. - The digest's live list for marshal::route leaves out shims::stale rows. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Castor-o7
force-pushed
the
ral-202-roster-order
branch
from
November 17, 2026 20:21
186d6be to
a868ae1
Compare
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
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.
RAL-202: the 7300 roster shows live sessions first and folds stale ones away.
Why: Josh, via the lamplighter and the watchman: the roster was "full of stale sessions, and the data is higgledy-piggledy". Read live, 7 of Marshal's roster rows are hook-only leftovers, and the table sorted every row by project.
The rule (
shims::stale) is one function, used by the viewer,stalls,wakeandbrief. A row is stale when:The stale reason names the fact and its label: "bound to no connection [recorded]" or "its pid is 0 [declared]".
Viewer:
data-keyso a changed count doesn't close it. Each row says why it's stale.last_activity_at, and "since: not reported by its roster" when it has no time. It still never says "for" an age taken from the last tool (RAL-124).stalls / wake / brief: stale rows are dropped before teams are matched, so a leftover is never taken for a team's live session. Their reports give
stale_sessions;rally stallslive givesstale_sessions: 7.Tests:
-D warningsare clean, and the fullcargo testpasses with browser tests (RALLY_TEST_CHROME): 580 lib tests and 54 web tests.Second commit, the watchman's follow-up to RAL-199 part 2:
marshal::recipientmatches a nickname only against live sessions: bound to a shim's connection, or holding a wait the daemon has now. When only a stale row holds the name, it says so.send_atwatches the wake watchers, so a waiting hook-only session can still be reached. The digest's live list formarshal::routeleaves out stale rows. So a ghost never ends a route early or makes a live nickname ambiguous. Unit tests cover a ghost sharing a live name, a ghost-only name, and a hook-only waiter. The ignored e2e still passes (3 of 3).🤖 Generated with Claude Code