Both packages import @deepseek-ai/dsh-llm at runtime (createUserMessage)
but declared it only in devDependencies; any resolver that honors
declared runtime dependencies resolves the import to a stale or missing
artifact. Found by the (since removed) source-launch declared-dependency
check.
Clearing waitingApprovals in handleConnected raced the reconnect replay:
mux frames flow from stream open while onConnected waits for the
readiness handshake, so a replayed approval/requested could land first
and be wiped — amber dot and answerable card lost until the next
generation. The sweep moves to generation death (onStateChange
'reconnecting'), before any next-generation frame can exist, and now
also drops buffered answerable frames (approval/question pairs) whose
dead-generation rpcIds could never be answered — a session instantiated
later no longer replays zombie takeover cards. session/queued buffering
already re-baselines per generation; this closes the same window for
the interaction frames.
Two races from the #572 review, still live in the ported registry:
An ask whose signal aborted between the service's own check and the
microtask-deferred waterfall dispatch would register its abort listener
AFTER the signal fired — never invoked, entry pending forever, zombie
frame on every mux replay. The answerer now settles 'cancelled'
synchronously before publishing anything.
The audit back-scan let a callId-less ask claim the newest unclaimed
asked record even when that record carried another call's id. Pairing is
now shape-symmetric: callId-bearing asks take exactly their call's
record, callId-less asks take only callId-less records — neither can
steal under parallel asks.
Disposability parity with the question provider: a gateway disposed while
approvals are pending settles every registry entry as 'cancelled' (the
service's fail-closed vocabulary), so no ctx.approval ask dangles past the
proxy's lifetime and mux subscribers see the withdrawal. Spec mounts the
proxy on its own fiber and drives dispose with a live ask.
Addresses the ds-review-bot suggestion on PR #851.
A popup on a host command is not a second command — it is what that
command's BARE invocation does on this client. CommandContribution loses
hostBacked (contributions are pure client commands again; a host-name
collision fails loud, unchanged for /model), and the contract gains
CommandDecoration + command.decorate(): key = the HOST command name, no
catalog row, no claim participation. Dispatch consults decorations only on
the bare paths (menu pick / bare enter) after the host row resolves; space
and argued enter never see them — the two edges hostBacked had to guard
explicitly hold by construction in the decoration model. A decorated name
with no host row in the session's directory never fires (a decoration
cannot manufacture a command).
ui-permission switches register→decorate with zero behavior change
(options still read the permissions projection; a pick still submits
'/permission <preset>'). Specs rewrite to the decoration semantics: no
catalog row, bare-enter popup vs argued-enter host claim, space host
claim, no-host-row miss, unavailable fall-through, duplicate fail-loud.
Selecting a preset from the hero pushed the session into the conversation
view: the /permission switch logs its command/run + command/done pair, the
pair folds into flow nodes, and the composerPhase predicate counted ANY
node as conversation — so the hero (composerPhase === 'blank') collapsed.
The host-side blank bit was already correct (sessionBlank = no turn/start;
knob events open no turn), but the client derives its phase from window
content, and command rows are log-only records, not conversation.
derivePhase's hasContent now excludes command nodes — the client mirror of
the host predicate. The knob events themselves never fold (not
surface-eligible), so the pair was the only leak. Covers /plan on the hero
identically (same lifecycle pair, same predicate).
Specs: the host blank spec pins the three knob events as standalone
events; a session spec drives the /permission pair through the live path
and asserts phase stays 'blank' while the command node renders.
TestWorkspaces.listDirectory now records the signal and passes it to the
installed stub, mirroring the production face, so cancellation
integration tests can observe or reject on a superseded scan instead of
the harness silently dropping it.
Supersession (newer navigation, path editing, closing, unmount) now
aborts the in-flight listing's request instead of only discarding its
result: the browser mints an AbortController per listing, the signal
rides the workspace face (IWorkspaces.listDirectory gains an optional
signal) onto the fetch carrier, and the Host scan stops with it (817's
cancellation chain). apps/cli keeps both picker packages as dependencies
so the documented one-row cordis.yml swap to the native backend resolves
at boot.
Same contract as the path editor: trim only rejects an all-whitespace
draft, and the Host receives the original spelling — the backend accepts
any non-blank single segment verbatim, so trimming here would create and
select a different sibling.
Trim now only detects a blank draft; the original text navigates — a
real directory name may end in whitespace, and trimming would list its
sibling or adopt the wrong workspace.
The aborted exit no longer awaits close (Node queues it behind any
in-flight read, chaining the departed caller back onto the very stall
the abort escaped) — the abandoned close's failure is swallowed, it has
no consumer. Symlink stat probes race the signal too, with a
per-candidate abort check between probes, so a stalled probe target
cannot keep a departed request alive. The deferred handle cleanup after
a lost opendir race now consumes its own close failure instead of
leaking it as an unhandled rejection.
The confirm/cancel pair wraps onto its own row when the viewport-clamped
card is too narrow for the whole footer, so Open stays visible instead
of clipping past the card's hidden overflow.
