Three P1 lifecycle defects and two P2s. The focused suite was 20/20 with
every one of them live, which is the part worth keeping.
**1. The crashed primary respawned forever underneath the fallback.**
Round 2 skipped the retire call for terminal servers to avoid corrupting
them — but the crash had already armed `next_restart_at`, and
`maybe_restart` fires on every elapsed backoff with no attempt ceiling.
The broken command kept respawning under the live fallback.
The right call depends on the state, and each is wrong for the other:
`forget` REQUIRES a terminal state and removes the client outright,
which also drops the restart timer; `stop` is for a live one and
corrupts a terminal one (its not-initialized branch parks it in
`ShuttingDown` forever). `retire_server` now dispatches on state.
**2. Re-attachment targeted whatever buffer was active when the
asynchronous verdict landed.** `_attach_buffer` is an active-buffer-only
seam, and "some attachment now names a different server" is satisfied by
an unrelated Rust buffer — clearing the retry and leaving the Lean buffer
stale forever. The initiating buffer is now captured and the retry waits
for it.
**3. A failing fallback retried every tick forever, silently**,
contradicting acceptance 27's promise that a second failure surfaces.
"Waiting for the old server to go" and "attempting the replacement" are
now separate: once the old one is terminal or gone, the replacement is
attempted EXACTLY once, and a spawn failure is reported.
**4. The Lake version parser was being applied to arbitrary wrappers.**
`version_below_3_1` encodes lake's output contract; a working
`my-lean-wrapper` reporting "wrapper 1.0" would have been replaced
despite its server initializing fine. The version probe is now gated on
the command's basename being `lake`. The FAILURE latch stays
command-agnostic — that one keys on the server actually not starting,
which is true of any command.
**5. An unconfigured Lean server was reported as a failure** and latched,
poisoning the session so a later configuration could never take effect.
Absent config or command now means disabled; only a configured command
that produced no attachment is a failure.
**6. The ledger recorded pre-fix counts** after the fixes were pushed.
Now 25/25 and 3,214. That is the #161 fmt-blocker error in a slower
form: verification must describe the pushed tree.
Sign-offs requested in review: `M.fallback` is now `M._fallback`, an
underscored test seam, and its idempotence check compares args as well as
command — the same command with different arguments is not "already
applied". Dropping the `command ~= "lake"` guard stands for the failure
latch only.
Five regression tests added, and **three of them were too weak on first
write; only bite-testing found it**:
* asserting "no live non-fallback server" misses a respawn loop,
because a respawning server sits in `crashed` most of the time —
`attempt` is the observable that counts respawns;
* returning to a buffer with `find_or_open` re-fires
`buffer.after-load`, which repairs the attachment regardless of the
code under test — `switch_buffer` is the honest return;
* a MISSING command fails synchronously inside `after-load` where the
rebuild happens inline, so the async race cannot occur — only the
probe path exercises it.
Each of the five now fails against the exact round-2 mutation it targets.
Round 1 review, four P1s. All real; the first two mean the fallback did
not work at all.
**1. The latch swapped the config but never spawned or re-attached.**
Nothing re-fires an attach on a config change and `attach_buffer`
early-returns for a live attachment, so the buffer stayed bound to the
server that had just been stopped. The user got a config edit and no
language server. `fire_latch` now rebuilds through a new
`pmacs.lsp._attach_buffer` export.
Two mechanics had to be right for that rebuild to happen at all:
* It is **retried on the tick**, because `pmacs.lsp.stop` leaves the
state `shutting-down`, which `server_is_live` counts as LIVE — an
inline re-attach early-returns the stale record and the swap is a
silent no-op.
* The latch **does not stop an already-terminal server**, and this is
a substrate bug worked around rather than a style choice.
`LspManager::stop` on a `Crashed` client takes its not-initialized
branch, terminates the dead process, and sets `ShuttingDown { ..
None }` on the premise that "the next exit observation cleans up" —
but the exit already happened, which is what made it `Crashed`. No
further event arrives, so the client is stuck in `ShuttingDown`
forever: `server_is_live` reads it as live so `attach_buffer` never
rebuilds, and `forget` refuses it for not being terminal. Stopping a
dead server is what makes it un-replaceable. Named in framing §6; the
fix belongs in `stop` and changes behavior for every language.
**2. A missing `lake` bypassed probe and latch entirely** — the single
most likely real failure. `ensure_server` swallows a synchronous ENOENT
and returns nil, so there was no attachment, and the hook keyed on
`active_attachment()` returned before arming anything. The hook now keys
on the buffer's LANGUAGE and treats a Lean buffer with no attachment as
the failure itself.
**3. `waitForDiagnostics` omitted `version`.** Lean's
`WaitForDiagnosticsParams` is `{ uri, version }` (v4.9.0,
`src/Lean/Data/Lsp/Extra.lean`); the request is how a client says which
revision it wants. It looked correct only because the fake server echoes
any payload — so the fake server now validates and returns InvalidParams
without it.
**4. The ledger stated the dangerous stacking order** in one sentence
and the correct rule in the next. Fixed to say BEFORE. A safety rule
written twice with opposite senses is worse than not written.
Also (P2): the probe/latch suite now drives the production path —
`buffer.after-load` -> ticks -> probe drain -> latch -> re-attach — with
real executable stubs, and asserts the originally opened buffer ends up
on a LIVE server. Round 1's acceptance 36 asserted every server was
terminal, i.e. pinned the ABSENCE of the fallback it claimed to test.
`M.fallback` is a table so the suite can point it at a working stand-in;
the probe now spawns `cfg.command --version` rather than a hardcoded
`lake`, which is also more correct for a user who configured a wrapper.
`swap_to_fallback`'s `command ~= "lake"` guard is gone: the latch fires
only when the configured server actually failed, one visible fallback
beats no server, and `probe.latched` is what keeps it to exactly one.
Three new bites, all against the committed tree: no re-attach after the
swap -> three latch tests fail; hook keyed on the attachment -> the
missing-`lake` case fails; `waitForDiagnostics` without `version` ->
acc37 fails with the server's InvalidParams.
Rev 5 said acceptance 34's second edge was a killed buffer. Implementing
it showed that is false: the Rust core fires exactly five hooks —
buffer.after-edit, buffer.after-load, buffer.after-switch,
frontend.detached, process.after-tick — and there is **no buffer-kill
hook**, so lsp.lua never tears an attachment down and the drain keeps
reaching that server. The premise (the drain builds its sid list from
`attachments`) was right; the inference needed attachments to be removed
on kill, and nothing removes them.
The reachable leak has the same root cause by a different path.
`attach_buffer` drops a sid from `attachments` the moment
`server_is_live` reports false and rebuilds against a fresh server — so
`crashed` / `stopped` is the event *least* likely to be drained, and an
event-driven purge leaks in exactly the case it exists for. The purge
therefore polls `pmacs.lsp.list()`, which enumerates the manager
directly. Acceptance 34's second half now exercises a server in **no**
attachment, which is the shape that discriminates: bitten, an
event-driven purge fails it while the attached case still passes.
§0.1 finding 6, Q#LN9, and acceptance 34 all updated; the wrong wording
is left visible with its correction rather than quietly replaced, since
the mistake is the useful part.
Ledger gains the Stage 3a lane: branch, worktree, what ships, both
corrected claims, the `install_async` load-order trap, the recorded
bites, the one knowingly unpinned guard, and gate results.
Two review findings, both revision edits.
**Q#LN8's marker test was wrong in the other direction.** Rev 5 fixed
the directory case by reading a byte and requiring a non-nil read — but
an **empty** `lean-toolchain` reads nil at EOF too, so that rule
declines a marker that exists, silently, falling through to
`pmacs.project.detect`. Marker semantics here are `lean4-mode`'s
`locate-dominating-file` semantics: existence, not content, and a
`lean-toolchain` can legitimately be empty.
The discriminator is `read`'s second return, probed on LuaJIT 2.1:
| Path | `io.open` | `f:read(1)` | Verdict |
|---|---|---|---|
| file with content | handle | `"l"`, no error | marker |
| empty file | handle | `nil`, no error | marker |
| directory | handle | `nil`, `"Is a directory"` | decline |
| missing | `nil` | — | decline |
So `local data, err = f:read(1)`, declining only on a non-nil `err`. The
rule needs no per-platform re-probe: both directory behaviors are
declines, since a platform whose `fopen` refuses a directory fails at
`io.open` and one that opens it fails at `read`. There is no platform on
which a directory both opens and yields a byte.
Acceptance gains **24b** (an empty `lean-toolchain` marks a root) beside
24a, with the obligation that each be shown to fail against the
implementation satisfying only the other. A suite carrying just one is
satisfied by a resolver silently wrong for the other case — which is
precisely how rev 5's first answer got written.
**Citation sweep.** Round 4 stated the `project_root_for` correction in
§0.1 without editing the citation in §2.5; the correction and the fix
are different acts, and noting one is not doing the other. Review caught
a second stale citation (`handle_server_requests` at :1448), which
prompted a sweep of every `file:line` from §2.4 onward. Four more were
stale. All six: `project_root_for` 513 → 592, `ensure_server` 527 → 610,
`handle_server_requests` 1448 → 1549, `take_typed_edit` 12798 → 12827,
`pair.lua` 213 → 229, `compile.lua` 264 → 266. Six others were verified
good and left alone, listed in §0.1 so the next sweep knows what has
already been checked.
Q#LN15's present-tense "the change is small and spans two files" now
reads as past tense with its PR number, since that stage landed. Its
pre-#161 line numbers stay as written — historical record, not
navigation.
Stages 1 and 2 landed (#160, #161). Re-scouting Stage 3 against `main`
@ `46a1b8f` — six merged PRs past the rev-4 snapshot — produced three
findings that change the plan and four that confirm it. Two were
established by running Lua in a fresh `EditorState` rather than by grep,
and are marked *probed* in §0.1.
**Stage 3 violated this document's own splitting rule.** §4 says "no PR
in this arc mixes a cross-cutting substrate change with Lean feature
content" and "a reviewer looking at Stage 3 sees only Lean" — while §4's
own risk column for Stage 3 read "two `lsp.lua` generalizations". Those
cannot both be true. One generalization shipped as Stage 2; the other is
Q#LN9's dispatch seams, which modify `handle_server_requests` —
confirmed the only production drain of LSP events, since
`LspManager::take_all_events` has no non-test caller. By the test that
justified splitting Stage 2 out, that is cross-cutting substrate. Stage
3 is now 3a (seams + canonicalizer, no Lean) and 3b (the Lean server),
strictly sequential.
**The Lean resolver could not satisfy the contract Stage 2 documented.**
#161 established that a configured root reaches `file_uri_for` verbatim
and that the resulting URI is the affinity key. Probed:
`pmacs.editor.file_path()` is not canonical — opening
`<tmp>/linkpkg/sub/./../sub/a.lean` through a symlink yields
`<tmp>/linkpkg/sub/a.lean`, lexical collapse only. No canonicalize
binding is exposed to Lua, and `pmacs.project.detect` canonicalizes but
returns nil without a marker. So one Lake package opened by two
spellings would spawn two `lake serve` processes — the bug Stage 2 was
built to prevent, re-entered through Stage 3's door. New Q#LN20 adds a
synchronous `pmacs.fs.canonicalize`; it rides 3a, and it serves every
future function-valued root rather than only Lean's. Two alternatives
are recorded with why they were rejected — the `detect`-anchored walk in
particular is incorrect, not merely inelegant.
**`pmacs.fs.stat` is unusable in the resolver.** It is async and the
resolver runs synchronously inside `ensure_server` ← `attach_buffer` ←
`buffer.after-load`, with no coroutine to await on. Probed: `io` and
`os` are exposed in the sandbox, so the marker walk uses `io.open` — the
opposite of what a reader would assume, hence Q#LN8 now says so. One
edge, also probed: `io.open` succeeds on a directory, so the walk reads
a byte rather than testing for a handle, and acceptance 24a bites the
version that does not.
Confirmed rather than changed: Q#LN7's stop-before-respawn is necessary
(default policy is OnCrash, the termination handler never consults the
exit code, and `maybe_restart` has no attempt ceiling — a broken `lake`
respawns forever; `stop()` setting `restart = Never` is what disarms
it); the response seam works as specified, since `Response` events are
pushed unconditionally and `send_request` returns the keying id.
One confirmation narrowed the design. `handle_server_requests` builds
its sid list from `attachments` and `push_event` is uncapped, so
subscribers fire only for servers with a live attachment. That turns
acceptance 34 into a reachable leak: killing the buffer with a request
outstanding strands the registration behind a drain that no longer runs.
The purge is now driven from both edges and 34 exercises the buffer-kill
path, which is the one a user can reach.
Also: §9 states the lane's coherence impact per COHERENCE §20 (journey
steps, interaction islands, config registry, background attribution),
including the honest note that 3b makes §2's step-3 grade marginally
worse by adding one more instance of the silent-spawn-failure class.
Three items are named in §6 rather than paid: the uncapped event queue,
the dropped `cfg.restart`, and surfacing the spawn failure itself.
Acceptance keeps every rev-4 number. The two split sections are
bulleted with literal labels because a markdown ordered list renumbers
from its first item, and 3b's criteria are non-contiguous; round 3's
finding 4 was stale references surviving a renumber, and not renumbering
is the cheaper way to not repeat it. Stale cross-references from the
split were reconciled in the same pass, and `project_root_for`'s
citation was corrected from 513 to 592 per COHERENCE §25.
The approved framing for Arc 8, revision 4, after three review rounds.
Seven stages: grammar/mode, multi-root LSP affinity, the Lean language
server, the Unicode input method, the goal view, the #eval output
channel, and module hierarchy. 19 decisions, 64 acceptance criteria.
Committed as this branch first commit per the house workflow; the
implementation of Stage 1 follows.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Levi Neuwirth