- status: done in the story frontmatter (was the non-conventional
"complete"; the board's axis uses done/in-progress/pending/hold)
- board row 11: what landed, the WO_WAL_UPDATE correction, the ceiling
removed as unreachable, and the one criterion still weaker than
written (expected value, not a resident: all oracle)
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit de39a88e81e97bf6f8b75e9f15331aae20f7ab29)
- flattened row image is WO_WAL_UPDATE, not WO_WAL_INSERT: the row's
original INSERT is already in a live log, so a second one for the same
id is a duplicate replay refuses as corruption. INSERT is right only
for compaction, which builds a fresh log
- remove WO_CKPT_MAX_GARBAGE: with the absolute term at 64 MiB, garbage
large enough to reach a 256 MiB ceiling has already tripped it, so the
branch was unreachable. Postgres needs both constants because it
thresholds on tuples with its pair at opposite ends; this thresholds
on bytes, where one constant does both jobs
- test_delta_chain_flattens_at_k: chain depth stays <= WO_DELTA_MAX_HOPS
across 2K+2 updates, and a reset is observed
- test_delta_chain_flatten_replays: a flattened chain replays correctly
- test_keys_resident_indexed_across_flatten: a delta on an indexed
column composes with flattening, checked at every step across the
bound and after restart. Found no product defect
- test_should_compact_absolute_and_ceiling: pins the absolute term, the
boundary just under it, and the small-log case the ratio still governs
- test_wal 5700 pass / 0 fail; wovm-test and woc-test green
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit f93b5d9db753305c297e868d977670e6d703684c)
TESTS DELIBERATELY HELD at the developer's instruction — logic only.
The existing suite passes (36 suites, 0 fail) but exercises NEITHER new
behaviour: nothing builds a 16-deep chain, and no checkpoint test uses a
log near 64 MiB. Green here means "did not break what existed".
- tier 1: wo_wal_fold_row_at gains hops_out. The walk already visits
every hop, so the depth is free — this is the design's pd_prune_xid,
a cheap "is work worth doing" hint taken from work already happening
- the update path branches on it: past WO_DELTA_MAX_HOPS (16) it writes
a full-row image instead of a delta, terminating the chain. `r`
already holds the complete post-update row because index maintenance
required folding it, so flattening costs bytes, not an extra read
- wo_wal_append_row_image encodes from a caller-held row, as
WO_WAL_INSERT: a chain's base must replay into a database where
nothing precedes it, so replay/compaction/fold need no change
- tier 2: should_compact gains a TRIGGERING absolute term and a ceiling.
Our `floor` SUPPRESSES on a small log — the opposite of postgres's
vac_base_thresh, which triggers on a small absolute problem the
proportion hides. We had the proportion and the suppressor and
neither real guard
- verified by construction, not test: both update entry points converge
on row_apply_field_keys; db.c captures next_offset BEFORE calling in,
so the re-point is transparent to which record type was written
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit 1b808abd5942de81c3a6416714d1302384103040)
- new `residency` leg in db-bench.py driving docs/examples/residency-bench:
two tables identical except the annotation, control cap + binding cap
- ITS OWN PROGRAM, not a db-bench mode: declaring a resident: keys table
is a WHOLE-PROGRAM constraint, so the no-WO_DATA refusal fires for
every mode in the module. Putting those classes in db-bench's shared
types made growth/ceiling/randread — which run without WO_DATA —
refuse to start. Caught by running the leg, not by reading it
- gates the RATIOS, waives the absolutes: ops/sec under a cap is swap
and disk I/O and belongs to the box. Same split randread makes
- rss_ratio 2.55 floor 2.0 tol 10% (structural, like bytes_per_row);
overcap_vs_swap_x 1.53 floor 1.0; in_ram_cost_x 4.23 ceiling 8.0;
all_collapse_x 105.4 floor 2.0
- all_collapse_x exists because the leg's FIRST run silently measured
nothing: at QUICK's 40k rows a 48 MiB cap binds neither mode, so the
"over-cap" half was not over cap. The cap now scales with N and the
leg asserts it binds
- verified the gate bites: rss_ratio 1.4, overcap_vs_swap_x 0.6 and
in_ram_cost_x 12.0 are all rejected
- task 7 closed: both criteria moved to Met with how each was verified
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit a310496664982c372f51b43113465eb8ad9e9fb5)
- hreadall/hreadkeys: the same resident A/B as wread_*, but ~2 KB per
row so a GB of data is reachable in a few hundred thousand inserts
- the insert path is fsync-bound at roughly 2 000 rows/s, so row COUNT
is the expensive axis and row SIZE is nearly free — 20k rows already
produce 38 MB
- NOT RUN: the GB-scale measurement was called off. These modes are
committed working and typechecking so the leg can be run later
without rebuilding it, not because a result exists
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit abc276ac39dd45a1052b6ae24d25aedb9eea3e1e)
- two tables identical except the annotation, 200k rows, 40k reads in
one key order, WAL on ext4 (not /tmp, which is tmpfs here and would
have put the log in RAM), rootless cgroup v2 cap
- WIDE shape, 2.55x smaller resident set: 34.4 MB vs 87.5 MB. That is
the real win and the thing the mode was built for
- under a 48 MB cap (between the two resident sets): keys 19635 ops/s
vs all 12854 — only 1.53x faster than letting the kernel swap
- degradation is far gentler though: all collapses 105x from its own
uncapped throughput, keys 16x
- costs 4.2x read throughput when memory is not tight, and writes are
markedly slower — the keys fill did not finish in 2 min where the
resident fill plus 40k reads did. No design doc had costed writes
- THE UNANTICIPATED FINDING: cgroup limits charge the PAGE CACHE, so
moving rows to a file does not escape a container memory limit. WAL
37 MB + RSS 34 MB cannot both live under a 48 MB cap, so every pread
reaches disk. The premise "the page cache will hold the hot rows"
fails in exactly the deployment this targets
- first attempt used Int-only rows and showed parity; recorded, because
drop_payload frees a field's VALUE and an Int's value is its inline
slot word, so that shape cannot benefit and would have condemned the
feature for the wrong reason
- verdict: keep it, to fit ~2.5x more data in given RAM — not to make
an over-capacity table fast
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit 7cba9b1174b0bf581314b3147e25cc49e6f49464)
- fixes a limitation iteration 2 shipped: compaction was supposed to
bound chain length, but wo_wal_should_compact triggers on a whole-log
byte ratio and cannot see one hot row's chain
- tier 1, flatten on update: the update path ALREADY folds the row for
index maintenance and the fold already walks hop by hop, so it reports
depth for free. Past a fixed K it writes a full row instead of a
delta. Read <= K+1 reads, replay O(K^2) per row. No format change, no
per-row RAM, no new trigger
- tier 2: our compaction policy has a proportional term and a
SUPPRESSOR misleadingly called a floor; postgres's floor TRIGGERS on
small absolute garbage. Add that term and a ceiling
- design read from .dev/reference/postgresql, not recalled:
heap_page_prune_opt gates on an O(1) on-page hint then page fullness
against Max(fillfactor, BLCKSZ/10); autovacuum uses base + scale *
reltuples clamped by a max (50, 0.2, 1e8). Neither thresholds on
new-bytes-versus-old-bytes
- K deliberately does NOT scale with table size: postgres scales a
table-level aggregate with proportional harm, ours is per-row with
additive cost, so scaling up would make big databases boot worst
- the story says plainly it should NOT be next: task 7 has still never
measured whether resident: keys beats the kernel's own paging
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit f667cad2cfbe187b5973440ab1af015b2df288f8)
- explains replay, the row chain, how a checkpoint flattens it, and why
replay of a long chain is quadratic
- worked SKU example with the actual record layout and back-pointers,
and a trace of the fold showing first-seen-wins
- states plainly why the checkpoint does not bound the hot-row case:
both triggers are ratios over the whole log and nothing counts
per-row chain length
- records the bounded-memory vs linear-time conflict behind the O(N^2)
replay rather than presenting it as an oversight
- closes with the reviewing lesson, since this shape survived several
rounds: complexity bugs hide in the caller's loop, not in the linear
helper being read
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit e6434403d566b5d72a24e9c0dcad1f25a2c16320)
- "no storage behind it" / "nothing yet stores a table that way" was
false — CRUD, checkpoint survival and updates all landed; replaced
with an accurate summary naming task 6/7 as what remains
- the three checked delete/delete-replay/update criteria sat in
Outstanding despite being done; moved to Met, leaving Outstanding
holding only genuine task 6/7 work
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit b575678ee95fa3125fa4e8145320a9cdca10ba38)
- loader.c: delete the INCOMPLETE-update BAIL; durable:false +
resident:keys stays refused (nowhere to read from)
- table.c: root-cause fix for the Text-index gap — a keys-resident
borrow now holds ENGINE values, matching wo_row_ptr's contract
(table.h's "no VM pointer" doctrine), not a VM-decoded row. Fixes
idx_hash/idx_cols_equal/wo_idx_probe AND db.c's GET_FIELD/PROBE
arms with one change; reproduced pre-fix as an ASan
heap-buffer-overflow
- docs/examples/residency: Product is genuinely resident:keys;
residency-accept.sh's refusal leg replaced by proving the program
runs and stock survives a restart (11/0)
- test_wal.c: oracle test drives resident:all and resident:keys
through the same update sequence and asserts identical rows;
Text-indexed-update test catches the representation bug; five
pre-existing tests corrected to the fixed contract (4746/0)
- story, README, status board, CODE-LOGIC.md updated; three known
limitations documented: mid-drain stale reads, O(N^2) replay in
chain length, compaction blind to per-row chain length
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit b87c68f950f01aa5e572fbb86a0f374adc83d813)
- the plan placed the fold near wo_wal_read_row_at at "line ~600";
it is at line 794
- every other citation verified: wal.h:46, table.c:452/487,
db.c:88/289, wal.c:738, wal.c:861
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit c6cd48679e21e510696391b340c735aa0ad098f9)
- 6 tasks: the record kind, the fold, updates + indexes, the request
path and group commit, replay + compaction, then lifting the loader
refusal and proving it end to end
- the fold is written ONCE and called from three places; the tests are
arranged to prove each caller separately, and the plan says that
wanting a second fold "just for this caller" means the design failed
- Task 2 includes a same-field ordering test specifically, because a
fold walking the chain backwards the wrong way returns plausible data
and is otherwise invisible
- Task 6 step 1 audits the db.c request arms BEFORE lifting anything —
they were never audited for keys-residency the way the inline path
was, and the last audit of that kind found delete corrupting memory
- self-review found the spec's crash criterion had no task: added a step
that commits a delta, skips the re-point, and replays, which is the
state a crash between barrier and flush leaves behind
- every symbol the plan names verified to exist in database/src
- code-free per house convention; the writing-plans skill wants code
blocks and the project rule overrides it
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit abb8fc9454bdca9d8d703c8cfeb224f89a93d14d)
- updates append a DELTA (class, id, field, value, back-pointer), not a
full row. The workload decides it: a product catalogue changes one
narrow field of a wide row on every order, so a full-row append would
rewrite every field to move one integer on a shop's hottest path
- the back-pointer keeps the id map at one slot per row, which is the
mode's whole premise; a map growing per update would defeat it
- ONE fold function, three callers (read, replay, compaction). Three
implementations of one rule is how they drift, and a fold that differs
between reading and replaying is a database that changes its mind at
boot. Named as the design's principal risk
- indexed columns MAY change: price is exactly what a catalogue indexes,
so forbidding it would be a restriction users meet immediately
- no chain cap, deliberately. Compaction already rewrites live rows, so
every checkpoint resets every chain, and deltas grow the log which
pulls the next checkpoint forward — the workload that lengthens chains
triggers the fold that shortens them
- the risk that accepts: one hot SKU under an otherwise quiet write
rate. Task 7's benchmark must include it
- supersedes the 2026-08-26 spec's one-line full-row Update sketch,
marked in place rather than left as a second design in the tree
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit c9a88b05c4c62a5df13253686c990aaccc3f9cbb)
- the motivating workload was an audit log, which is append-only and so
argues for nothing. A catalogue is the real case: stable ids, and
stock moving on every order while name/price/sku sit still
- Product (durable, resident: all) and Cart (durable: false), with
place_order decrementing stock through a write-through field assign
- run `order` twice and stock goes 10 -> 7 -> 4: a level below the
seeded 10 can only mean an earlier order's UPDATE replayed. That is
the stronger claim — not just that inserts survive, but that a field
change does
- caught by running it three times: my first assertion required
before == 10, which only holds on a fresh seed and failed on the
third run even though the data was correct
- gate gains a leg for the update-replay claim; residency 12/0
- the commented resident: keys block now argues the DESIGN too: only
stock changes per sale, so appending the whole row would rewrite
every field to move one integer on a shop's hottest write path
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit d4104dcc5e3a460459dbf2d24043ce25b113bcdf)
- wo_row_remove's keys arm borrows the row from the log to find its
index entries, and a borrow reads through db->rt->wal. At boot that
pointer is not wired yet: main.c replays first (main.c:226) and
assigns rt.wal afterwards (main.c:268)
- so the borrow found no log, the remove failed, and replay reported a
valid tombstone as CORRUPTION. An UPDATE record would have failed the
same way, since replay applies it as remove-then-recreate
- replay now lends the runtime a read-only view over the fd it already
has open, for the replay's duration only, and restores what was there
- broken by the delete fix in 76b8fd9 — deletes worked in-process but
their tombstones broke the next boot. Unreachable in production only
because the loader still refuses the annotation
- pinned by test_keys_resident_delete_then_replay, verified failing
against the unfixed code (2 failures) and clean with it
- found by asking whether the read-modify-append plan was ready, not by
a gate
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit dc25462461b9f79d70c803f7174adc90fa16c90e)
- docs/examples/residency: one program, two tables filled by the same
loop, differing only in the annotation. Run twice against one WO_DATA
and orders replay while sessions do not
- the example checks its own claim (exits 1 if a durable:false table
survives, or a durable:true one fails to replay) rather than narrating
it in a print
- resident: keys is written out as a commented block with the loader's
exact refusal, so the frontier is visible in the example rather than
only in a story. It documents WHERE the refusal happens: woc compiles
it and emits a .wob; wovm exits 2, because the annotation is a
load-time property
- residency-accept gains two legs: the example runs and its restart
claim holds, and the refusal message the README quotes is checked so
doc and code cannot drift apart
- the gate writes the example's output to /tmp/residency.log,
banner-separated, for tail -F
- README commands verified verbatim; they needed mkdir -p because wovm
will not create WO_DATA
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit c9c7e03e62c3918cb65ef7994d1e33a0c5337b71)
- wo_row_remove read the id map's value as a slot, but on a keys table
that value is a LOG OFFSET (hput(t, id, wal_off + 1)). slot_row does
no bounds check, so a delete indexed t->slabs[] with a byte offset and
then called db_val_free on whatever it landed on — arbitrary frees,
not a wrong answer
- keys tables now take their own arm: no slab slot, no bitmap bit, no
free-list entry to return. The index hook needs the row's values, so
the row is borrowed from the log for exactly that long
- wo_row_ptr carried the same trap and is public. It cannot refuse keys
tables outright (insert legitimately calls it while the map still
holds a slot), so it now detects the offset case — index past the
slabs, or bitmap bit clear — and returns NULL. Callers all handle NULL
- test_keys_resident_delete pins it; it SEGVs against the old code,
verified by reverting the fix rather than assumed
- found while auditing every hget() reader before narrowing the loader
refusal to allow benchmarking. The refusal was justified in the docs
by "updates are unimplemented" while actually standing in front of
this too: a guard whose stated reason is narrower than its real one
gets removed by someone who believes the stated reason
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit 76b8fd944af9ed062467bdf9ab93c2e96dd198cf)
- chain is a cross-track field (1-6); only databasev2 3 and 4 carry it,
and iteration 2 — critical path, in-progress — carries none, so the
board's chain query cannot see it
- 00-story.md's ASCII graph draws 3 before 4; the chain field and the
history both say 4 first (4 part A 08-28, 3 08-29). Graph is wrong
- graph also contradicts its own prose on edge direction, and still
draws the 2-5-6 path the 2026-08-27 amendment retired
- iteration 3's hazard section is stale: it says nothing fails "because
iteration 2's storage half is unimplemented", which 5c/5d ended
- and it was incomplete: it named offsets going stale, but compaction
walked the bitmap, which a keys row has no bit in, so those rows
would have been dropped from the new log outright — data loss, not a
bad pointer, and offset-rebuilding would not have caught it
- coupling is now bidirectional: wal.c compaction calls iteration 2's
wo_row_next_id / wo_row_offset1 / wo_row_set_offset
- nothing in the track is blocked on anything else in the track
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit 2ecaf0c95f40c6d6c8cc2fde687c82d63cda1f7e)
- loader's resident:keys refusal said "rows are still fully resident"
and "until tasks 5c/5d land". Both false since f606fc9. Corrected to
name the real blocker: UPDATE needs read-modify-append
- databasev2 00-story: the sequence graph drew 2->3->4, which reads as
3 needing 2 and 4 needing 3. Both backwards, and it still drew the
2->5->6 path the 2026-08-27 amendment retired. Redrawn stating only
real dependencies, with 4 and 3 shown as composing rather than
ordered, and the execution order that actually happened
- databasev2 03: the hazard and its Outstanding entry both claimed
nothing fails "because iteration 2's storage half is unimplemented".
Marked discharged, and recorded that the hazard named only half the
danger — the bitmap walk would have dropped keys rows outright
- databasev2 06: pending -> hold (largely superseded, revisit only on
a measurement); dated its 5c/5d references
- porch 01: rewritten to the settled shape. readiness ready, status
in-progress, phases B and C marked superseded with why
- porch 01 claimed time.after "is still a reserved builtin id". False —
builtin 90, implemented. That claim is what made the iteration look
cheaper than it is
- porch README gains honest ledger rows for both features (partial,
being rebuilt), not shipped
- skill-catalog README pointed at a story path that moved tracks;
linkcheck now 0 broken
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit b3d8c403e1d19ac27ec966de85cb293e0765795c)
- wo_row_read and the @unique shadow probe go through borrow/release;
release runs before every exit, including wo_row_read's early return
- updates on a keys table refused explicitly in wo_row_update_field and
the slot variant: no slab slot to mutate, and writing the borrow's
scratch would discard the write silently. Needs read-modify-append
- compaction walked the bitmap, which a keys row has no bit in — every
such row would have been dropped from the new log. Now walks
wo_row_next_id and re-points each row to where it lands
- moves records byte-for-byte (copy_record) rather than decoding: a
borrowed row holds VM values, enc_val expects engine values, and ASan
caught that mismatch as a 4294967292-byte memcpy
- wo_row_set_offset updates a value in place and never rehashes, so a
wo_row_next_id cursor stays valid while compaction re-points
- a compaction that fails after moving rows is fatal: the map would name
an unlinked temp file, and the intact log replays correctly
- test_keys_resident_survives_compaction pins both failure modes; rows
rewrite in hash order so offsets really move
- loader still refuses resident: keys — updates are not implemented
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit f606fc9b76d983cac2b348f03f7d4f01433cd905)
The branch was 17 ahead / 25 behind with 11 conflicting files, and drifting
further: db.c had been rewritten twice on master since (group commit, then
compaction). Resolved rather than rebased so both histories stay legible.
Conflicts, and how each was settled:
- db.c: BOTH semantics kept. Master's fatal path and compaction check now sit
behind the branch's `table_is_durable` predicate, in all three inline arms —
a volatile table reaches neither the barrier nor the compaction check
- db-bench sample: every mode from both sides (growth, growth-verify, randread,
replayseed, wmix) and ONE `boot` mode, which both sides had added
independently
- db-bench.py: all six legs kept. Both sides had also grown the same
WAL-size helper under different names; collapsed into one
- perf-targets: the branch's §5 (RAM ceiling) then master's §6/§7 — master's
numbering had already assumed a §5 it did not have
- story frontmatter: master's `status` (the landing truth) plus the branch's
`readiness` axis. 03 would have read `done` + `refine`, which is a
contradiction — it was brainstormed and landed on master, so `ready`
- board: both standup blocks newest-first; master's chain rows (a superset);
the branch's databasev2 1-2 rows with master's 3-4. Fixed a stray `|` in
master's row 3
- baseline: master's, then REGENERATED from a full campaign — 143 metrics,
132 checks, 0 failures with both sides' legs present
TWO HALF-EXPOSED FEATURES FIXED, because the merge rule is that master gets
no feature that is honoured in name only:
- `resident: keys` PARSED, set a .wob flag, and did nothing: rows stayed fully
resident. A developer could declare a 120 GB table keys-resident, watch it
compile, and be OOM-killed. The loader now REFUSES it with a message naming
what to write instead, until tasks 5c/5d land. The compiler still parses it
and its AST golden still passes, so the grammar work stays tested
- `durable: false` was honoured ONLY on the inline path. wo_db_exec_req had no
guard at all, so a volatile table written from an actor on a worker shard
would still be logged — precisely porch's session-table case, and precisely
what iteration 2 exists to provide. All three request-path arms now carry the
same predicate. Found by reading the merged code, not by a test: the obvious
probe runs main() on the primary and therefore only exercises the inline path
Verified on the merged tree: wovm-test 0, woc-test 0, oop-e2e 122/0,
residency-accept 8/0, db-bench 132/0, linkcheck clean.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
databasev2 3, task 6. Documentation, plus three gate-tolerance
corrections that are justified rather than silent.
- 04-db-binding.md: the NORMATIVE rule — compaction may run only where
nothing is staged (a correctness requirement, not scheduling), recovery
is unchanged, and a failed compaction is a missed optimisation rather
than a durability event
- database/src/CODE-LOGIC.md: why one file and not snapshot-plus-tail
(Postgres CANNOT compact — page deltas; ours are full row images, so a
compacted log IS a store), why rename is the whole crash-safety story,
why the dump flushes but does NOT fsync when it does, why the
replacement is preallocated, and where the trigger is checked
- README: the checkpoint knobs, the extended walstats line, the boot mode
- story -> status: done, with criteria split met/outstanding
- board: standup entry in the six-question shape, both rows rewritten
THE OBLIGATION IS AT THE COMPACTOR, not only in a spec: compaction moves
every record, so it invalidates every WAL offset iteration 2's
`resident: keys` stores, and the loop that knows each record's new
position must rebuild that map. Nothing fails today because that storage
half is unimplemented — it would fail later, looking like corruption.
Board claim corrected before it shipped: I wrote that the concurrency
chain is "complete". It is not — chain 5 stays in-progress because
databasev2 4's part B was never done and its premise was invalidated by
part A. Every link has landed its PLANNED work; that is a different
statement.
Gate tolerances, each with the measurement that justifies it:
- ckpt.pause_us_max is no longer gated relatively. The raw pause scales
with the live set and this workload's live set is not fixed (wmix's
hist_dump inserts a row per latency bucket), so gating it gates the
box. Added ckpt.pause_us_per_mb — the engine's own rate, gated for
real, and the metric that would have caught the 8x dump regression —
with the absolute 50ms budget still guarding the raw pause
- ram.*.msgrate 15% -> 70%. PRE-EXISTING, and measured: 10.7M-17.9M
msgs/sec across ten full runs, several predating this work — a 1.67x
spread against a 15% gate
- durable.sN.*.p99us 100% -> 300%, with more evidence than the first
widening: mixread 1043/2318/4147us, mixwrite 1623/4446us on the same
build. Floors stay the real guard and are not slack
Battery: wovm-test 36 suites 0 fail, woc-test, oop-e2e 119/0,
db-bench 117 checks 0 failures, linkcheck clean.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
databasev2 3, task 5.
Full campaign, same workload twice, differing only in whether
checkpointing may fire:
- WAL used 1962358 -> 907094 bytes (2.16x reclaimed)
- boot 114 -> 64 ms (1.78x), median of 3
- stop-the-world pause max 2651us against a STATED 50ms budget
The budget is asserted, not assumed: 50ms is a stall a serving process
can absorb without a client seeing a timeout, and the leg fails if it is
exceeded. The pause is O(live rows) — at ~181 MB/s a 1GB live set implies
~5.5s, which is the number an incremental design must be bought against.
The spec deliberately did not buy it in advance.
FOUND BY MEASURING: the dump was 8x slower than it needed to be. It
flushed through wo_wal_commit, which fdatasyncs, so it paid one barrier
per 256 records. Intermediate durability there is worthless — the temp is
not authoritative until the rename and is fsynced once immediately before
it. With a single final barrier:
- ~107KB live: 23948us -> 2903us
- ~500KB live: 36361us -> 7526us
- ~1.98MB live: 107649us -> 13212us
- marginal ~22 MB/s -> ~181 MB/s, sync-bound to bandwidth-bound
Correctness re-proven after that change: wovm-test 36 suites 0 fail,
test_wal 760 pass including the 40-round kill-during-compaction battery.
Two measurement defects of my own, fixed rather than reported:
- boot measured through the driver's run() helper reported 251ms both
with and without checkpointing — run() samples RSS on a 250ms poll, so
every timing floors at the quantum. Measured directly instead, median
of 3
- ckpt.reclaim_x was recorded as lower-is-better by the default detector,
which would have PASSED "reclaimed nothing" and FAILED an improvement:
the feature's central claim, gated backwards. Now higher-is-better,
gated at 15% while the wall-clock metrics stay wide — waiving them all
would have left the leg ungated, part A's task 4 mistake
- sample gains a `boot` mode that does nothing, so boot time is boot time
- walstats now reports compactions, pause max/total and compacted bytes
- baseline refreshed from the FULL campaign (N=20000, crash_reps=3), and
a fresh full run passes 116 checks 0 failures
- gate bites: reclaim_x doctored to 1.0 -> FAIL on exactly that metric
One flake seen and checked, not papered over: durable.sN.query.ops_sec
failed once at 53% below baseline. It is a read-only metric that touches
no WAL code, and a re-run passed 116/0 with the box at load 1.85.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Plan for the approved spec. Code-free per the repo convention
(docs/plan/discarded.md:54); the executor writes the code.
- T1 wo_wal_compact: walk live rows via the bitmap, append one INSERT
each through the EXISTING append path, fsync, rename over the live log,
fsync the parent dir, reopen the descriptor. Test asserts BOTH that the
log shrank AND that a replay reproduces the same rows/ids/values —
shorter alone is worthless, a truncating bug also passes that
- T2 a stale temp file is removed at open and never read. The test uses
PLAUSIBLE records, not garbage: garbage would be rejected anyway and
would prove nothing
- T3 the trigger as a PURE decision (used bytes, last compaction's
measured output, floor) so it is unit-testable without a store; env
knobs for floor and ratio, which is what makes the policy testable at
all. No timer, with the reason. The check is called only where nothing
is staged, asserted by a test that stages and expects deferral
- T4 kill -9 DURING compaction, extending the existing fork-based crash
battery. Asserts the PROPERTY — the store equals the pre- or the
post-compaction content, never a mixture, and every acked id survives.
Run repeatedly and state the count: it is a race, one green run proves
little
- T5 measure space reclaimed, boot before/after, and the stop-the-world
PAUSE against a stated budget. If the pause exceeds it, stop and report
— the alternatives are bought against that number, not before it
- T6 closeout, including the normative ordering rule in 04-db-binding.md
Constraints carried from the spec into every task:
- recovery must NOT change; a task editing the replay path should stop
- the dump must FLUSH PERIODICALLY. stage() grows the staging buffer by
doubling, so dumping a whole store through one buffer would hold the
entire store in RAM — the unbounded growth databasev2 1 identified as
how this engine dies
- a FAILED compaction is a missed optimisation, not a durability event,
so it must not take databasev2 4's fatal path
- gate tolerances must not be waived wholesale (part A's T4 made that
mistake), and the baseline is full-mode — writing a quick-mode baseline
over it is a regression part A also made
Deliberately NOT a task: rebuilding the `resident: keys` offset map. It
cannot be implemented against a feature that does not exist yet, so T6
records it as an obligation at the compactor and in the story instead of
a stub nobody can test.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
databasev2 3, chain 6. Brainstormed 2026-08-28 after databasev2 4 part A
landed.
Design: compact the log by rewriting it as one record per live row into a
temp file, fsync, rename over the live WAL, fsync the parent dir, reopen.
Recovery is COMPLETELY UNCHANGED — boot still opens one file and replays
it — and the crash criterion ("the same store as if the checkpoint had
never started") is satisfied by rename, not by code we must get right.
Read .dev/reference/postgresql for this. The finding is that PG's design
is UNAVAILABLE to us, which is what makes the simpler option legitimate:
- PG never compacts its WAL; segments before the redo point are recycled
by rename or unlinked. Its records are page deltas, so a compacted redo
log is not a store — hence heap files, a control file, a redo pointer,
a second recovery source and a separate process
- ours are FULL ROW IMAGES (apply_record implements UPDATE as
remove-then-recreate), so a compacted log IS a complete store. That one
difference deletes all of the above from the design
- what IS worth porting is the ordering discipline: publish the new
"recovery starts here" atomically and LAST, so a crash falls back. PG
needs a start-of-checkpoint redo pointer plus an end-of-checkpoint
control file update; we get the same property from one rename, because
we can swap the whole data set atomically and PG cannot
Forks settled:
- no snapshot format — the compacted log is the snapshot, existing grammar,
so no new encoder or decoder and the dump reuses wo_wal_append_insert
- one source, not two
- volume-only trigger, as a ratio against the LAST compaction's measured
output (the denominator is known exactly; estimating the live set would
mean estimating Text) with an absolute floor. NO TIMER — PG's exists to
bound loss from unflushed buffers and we have none; an idle log does not
grow. Copying the mechanism without the reason was the trap
- stop-the-world, with the pause measured against a stated budget rather
than assumed acceptable; alternatives are bought against a number
- compaction may run ONLY where nothing is staged (right after a barrier),
or a staged record lands in a file about to be replaced. Normative
Recorded before it can be found late: compaction invalidates every WAL
offset iteration 2's `resident: keys` stores, so the compactor rebuilds the
offset map as it writes. Nothing breaks today because that storage half is
unimplemented — it would break later, looking like corruption.
Also corrected exploration/postgresql/buffer-and-checkpoint.md, which was
wrong on two counts: PG does NOT update its control file by rename (in-place
full-block write + CRC32C), and its checkpoint sketch assumes writeonce has
segment files, which it does not and deliberately will not.
Grounding measured on master: seed 20000 leaves a 986614-byte log; 20000
updates take it to 2590262 bytes with the SAME live rows, and boot+verify on
that store is 155ms.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
databasev2 4 part A, task 6. Mostly documentation, plus one real fix the
full battery caught.
THE FIX. The drain held EVERY DB reply until the barrier — including
reads, which stage nothing and have no stake in durability. That parked
readers behind an fsync for no reason: durable.sN.mixread.p99 rose from
~1043us to 4057us. Only a statement that actually staged a record now has
its reply held. Caught by the gate, not by review.
THE TRADE, recorded rather than smoothed over. What remains is inherent: a
barrier blocks the owner shard LONGER (more records per fsync) though LESS
OFTEN, so anything queued behind one waits. Three full runs of the same
build gave durable.sN.mixread.p99 of 1043 / 2318 / 4147us and wmix.p99 of
8758 / 20000us — a 2-4x spread with the box near idle. So part A buys ~3x
write throughput at the cost of a longer, noisier tail on the owner shard,
and that is the strongest argument for part B (submit and keep serving).
- durable.sN.*.p99us tolerance widened to 100% WITH the reason in the
code: a 2-4x-variable tail gated at 50% gates the disk, not the engine.
The floor is the real guard and is not slack — mixread's (4172us) came
within 25us of tripping on the worst run. Baseline refreshed; a fresh
full run then passed 106 checks 0 failures
EXIT STATUS MOVED 3 -> 74 (sysexits EX_IOERR). 3 and 4 are already used by
SAMPLES for their own meanings — db-bench's own `verify` exits 3 on a
checksum mismatch, and it is the gate that exercises durability, so a
durability abort exiting 3 would have been indistinguishable from the
mismatch it should help diagnose. The low range belongs to programs.
Docs:
- story: progress, the payoff measured two ways, the cost side, criteria
split met/outstanding, and a "part B — its premise changed" section:
it was justified by "close the 66x gap", but that gap is two problems
and only the concurrent one was a batching problem
- board: standup entry in the six-question shape; both databasev2 4 rows
rewritten. They had said "close the 66x gap" — recorded as MIS-STATED
rather than quietly renumbered
- 00-wob-format.md and 04-db-binding.md: the normative failure contract
("a failed WAL commit traps WO_T_IO after un-applying the row") was
false; corrected, along with the tick-scoped group commit that never
happened
- database/src/CODE-LOGIC.md: where the barrier runs and why there, why
replies are held, why the inline path is asymmetric, the one failure
rule, and how to measure it
- db-bench README: the wmix mode, the env knobs, and the tmpfs warning
Battery: wovm-test 36 suites 0 fail, woc-test, oop-e2e 119/0,
db-bench 106/0, linkcheck clean.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
databasev2 4 part A, task 5.
Controlled before/after — same machine, same workload (wmix 4000 32),
same build except db.c and vm.c, two runs each interleaved:
- per-statement barrier: 2213 / 2177 ops/sec, p50 7183 / 7251us
- group commit: 6216 / 6525 ops/sec, p50 3458 / 3444us
- ~2.9x throughput, ~2.1x lower p50
The full campaign confirms it a second way: s1 takes the inline path and
commits per statement BY DESIGN, so within one build the shard configs are
batching-off vs batching-on — 1467 -> 5117 ops/sec, mean batch 1.0 -> 5.43,
peak 1 -> 57. 3.5x, agreeing with the 2.9x above.
Recorded honestly:
- the BEFORE p99 is at the histogram ceiling (hist_add clamps at 20000us
and both runs pinned there), so the true figure is >=20ms and unknown.
The improvement is AT LEAST 2.3x; the old p99 was off the instrument
- durable.sN.mixwrite went 480 -> 492 ops/sec, i.e. UNCHANGED. That was
the spec's original payoff metric and correcting it was part of the
brainstorm: mix performs 20 writes at C=4, mean batch 1.01. A workload
that never has two writes in flight cannot be helped by batching them
- seed is likewise unchanged: a serial writer has nothing to batch with
- so the payoff is real but CONDITIONAL — it appears where concurrent
durable writes fan into the owner shard, and nowhere else
Two traps recorded in perf-targets §6:
- do not benchmark durability on /tmp: it is tmpfs here, where fdatasync
is free. The same run reported 195000 ops/sec at p50 1us there against
2200 at p50 7200us on ext4 — no barrier to amortise, so the measurement
measures nothing. db-bench keeps its stores under bench/ for this reason
- the record count is not the update count: 7755 records for 4000 updates,
because hist_dump and the done-marker are themselves durable inserts
- FIXED a regression I introduced in T4: master's committed baseline is
FULL mode (N=20000, crash_reps=3, msg_n=200000) and I had overwritten it
with quick-mode values. Regenerated from a full campaign; the full run
now passes 106 checks 0 failures against it
- gate still bites: sN wmix ops_sec -70% -> FAIL on exactly that metric
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
databasev2 4 part A, task 4. Scope extended with developer approval: the
plan authorised touching the sample only for observability, but no
existing leg has enough concurrent durable writes to exercise group
commit at all, so the payoff was unevaluable either way.
The finding that forced it:
- `mix` writes on one op in ten with C=4 (all_mode calls mix_mode(n/10,
4); Mixer writes on i % 10 == 9), so the quick run performs 20 writes
total. Measured mean batch 1.01 over 3112 barriers, peak 3
- that is a property of the WORKLOAD, not the mechanism: peak 3 of a
possible 4 shows batches form whenever writes actually coincide
- `wmix N C` added: every op a durable write, C at once. Updates rather
than inserts, so it is comparable to mixwrite and the row count stays
flat. Histogram kind 2 — a replayed store still holds the seeding run's
kind-0/1 Hist rows and merging those would report someone else's
latencies
- WO_WAL_STATS=1 prints one line at exit: batches, records, peak_batch,
peak_staged. Opt-in, because it would otherwise pollute every durable
program's output. Counters live in wo_wal; no builtin, the numbers are
diagnostic and not part of the language
Measured, and it scales with concurrency exactly as designed:
- C = 4 / 16 / 64 -> mean batch 1.13 / 1.76 / 5.35, peak 3 / 10 / 39
- the gate's own legs: durable.s1 5412 records over 5412 barriers (mean
1.0, peak 1 — the inline path, one barrier per statement BY DESIGN),
durable.sN 7757 over 2296 (mean 3.38, peak 28) at 2x the throughput
- peak staged 1372 B settles the no-cap decision with a number: the batch
is tiny, so the upstream mailbox bound is sufficient
- mean_batch/peak_batch are higher-is-better (the default detector would
have called bigger batches worse)
- only the batch SHAPE metrics are waived to 100%; wmix throughput and
latency keep real tolerances (15% s1, 50% sN) — a blanket waiver would
have left the entire new leg ungated
- the live assertion `mean > 1.0` on the sN leg is what catches inertness
- gate bites: sN wmix ops_sec -60% -> FAIL on exactly that metric, 1 of 86
Verified: db-bench-quick 89 checks 0 failures; baseline refreshed (86
metrics).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Plan for the approved spec. Code-free per the repo convention
(docs/plan/discarded.md:54); the executor writes the code.
- T1 a failed barrier is detected and fatal — one entry point that names
the operation, errno, WAL path and batch size, then exits. The abort
path itself stays unexercised and the task says so rather than buying
coverage with a fault-injection switch
- T2 the barrier moves to the drain point and replies are held; the
request path stops committing per append. Riskiest task, and its risk
is one place: the crash legs. Plan says STOP if they fail, do not
adjust the test
- T3 the inline path takes the same fatal rule but keeps its own barrier,
with a comment explaining the asymmetry so the next reader does not
"fix" it. Looks like a no-op; without it the two paths disagree, which
is the unevenness the spec exists to remove
- T4 prove batches actually form BEFORE measuring the payoff — otherwise
a win gets attributed to the wrong cause. Also records peak staged
bytes, settling the no-cap decision with a number
- T5 measure, gate, write it down. If the payoff is absent, say so and
stop: part B must not start on an unproven premise
- T6 closeout, including the error catalogue — WO_T_IO leaving the write
path is language-visible and must be written down
Spec corrected while planning: it pointed at durable.s1.seed as the
payoff. Wrong, structurally — worker shards hold no WAL, so a queue only
exists when other shards write, and a serial writer has nothing to batch
with. The real target is durable.sN.mixwrite: 480 ops/s at p99 5888us
against s1's 1023 at p99 664, so adding shards currently makes durable
writing WORSE. That inversion is a better argument for the iteration than
the one the story recorded.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Brainstormed 2026-08-28. The iteration is split: part A batches, part B
(io_uring submission) is deferred until A's measurement says whether the
blocking boundary still dominates.
The story's premise needed correcting first:
- it says "replace fsync-per-commit with io_uring group-commit", but the
engine commits per STATEMENT — db.c calls wo_wal_commit right after
every append, all six sites, so each row change is one pwrite + one
fdatasync
- so two independent wins were being carried as one, and only the second
needs io_uring. The staging buffer already holds any number of records;
today it never holds more than one. Part A is mostly deleting calls
- iteration 22's numbers say A is where the payoff is: durable writes
4460 ops/s, mixwrite 1023 ops/s p99 664us, against 1.28M ops/s reads
Forks settled:
- batch boundary is QUEUE-DRAIN, not the tick this story had recorded: a
tick adds latency to a lone writer, taxing an idle system to serve a
busy one. Queue-drain self-tunes and needs no knob
- shard 0 holds each reply envelope instead of sending it, commits once
when the queue empties, then releases all — so a writer is acked after
the barrier carrying ITS record, which today is true only because
every batch has one member
- a failure between "RAM mutated" and "record durable" is a FATAL,
diagnosed abort. This replaces uneven behaviour that already exists:
insert rolls back, update and delete do not and say so in a comment
("RAM ahead of disk"). Batching would have multiplied that
- consequence stated, not slipped in: WO_T_IO leaves the write path
- no batch cap initially; peak staged bytes is measured so the question
is settled by a number
One gap disclosed rather than hidden: forcing a real fdatasync failure
needs mount privileges, so the unit test proves the error is DETECTED and
the abort itself stays covered by inspection.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Iteration 24 closes, absorbing 31 and 34. No code in this commit.
- stories 24, 31, 34 -> `status: done`, each with a landing banner. 24's
records the gate numbers and BOTH disclosed deviations: monitor takes
three arguments (the caller may be `main`, which has no mailbox) and a
v1 `call` reply is a typed scalar (which is what let the agreement be
checked at compile time, WO-E226). 31's notes it landed INSIDE 24 and
that a fifth mechanism it never anticipated came out of proving the
gate — the drain guarantee (40). 34's names the gap it did NOT close:
still no RNG, so CSRF/sessions stay blocked
- board: in-progress row cleared, marker doc deleted (convention), the
standup entry in the six-question shape, chain note — next link is
databasev2 4 (io_uring group-commit, chain 5)
- graph: PUBSUB2 (pub/sub + WebSockets, "rejected until here") -> done
- porch ledger: a WebSocket/pub-sub row added; the cancellation row now
says what it actually waits on rather than repeating "the arc"; the
README's "no WebSockets/SSE" limitation was stale — WebSockets are
supported, SSE and chunked encoding are not
- CODE-LOGIC: runtime/src gains the actor-lifecycle section (call, death,
the cap counter's sender/home-thread split, the monitor walk, the timer
list), the drain guarantee, and the digest section; docs/examples/chat
gains its own — actor topology, WHY two actors per connection, fd
ownership, and the shutdown choreography
Battery after the doc edits: wovm-test 36 suites 0 fail, woc-test exit 0,
oop-e2e 119/0, chat 11/0, web-app 46/0, linkcheck clean.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
- narrative said "five of ten tasks landed" and named the branch as the
live location; T4/T5 (ids 89/90) had landed and the slice merged to
master 2026-08-27 (60414a1, fast-forward)
- records what was verified ON master: chat 11/0 at the full 1000-client
soak, runtime 36 suites 0 fail, compiler 556 checks, corpus 119 checks
- T10 closeout is what still holds stories 24/31/34 open
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A message sent before the stop flag is observed must be delivered and run
before the engine stops. One rule; a spin count could never express it.
- root cause in `shard_main` (runtime/src/vm.c): NEXT_RUNNABLE() already
stated the contract — "a WORKER on stop keeps DRAINING ... so queued
shutdown messages (close frames!) still run" — but the IDLE branch
contradicted it, calling fib_reap_all and breaking on WO_IO_STOP,
abandoning its inbox for wo_engine_stop() to free wholesale
- an actor between messages is exactly that idle case, which is why a WARM
soak server hid it: warm shards held live fibers and took the right path
- fix: while the primary's drain window is open, an idle worker adopts its
inbox and runs what arrives; sched_yield on an empty poll so a drain
cannot burn a core per shard and starve the actors it exists to let run
- unreachable at WO_SHARDS=1: wo_engine_stop returns early at nshards <= 1
Measured:
- fresh-server SIGTERM drain: 5 of 16 failing before, 20 of 20 clean after
- `just chat` at the FULL 1000-client soak: 11 checks, 0 failures, both
WO_IO backends, ASan clean with zero leaks
- the fd leg settled at scale too: 1000 connections left the count at 44,
unchanged after 20 more — lazy per-shard init, not a leak
- runtime battery 36 suites (18 x both dispatch flavors) 0 fail;
compiler 556 checks 0 fail
- story: docs/stories/language-runtime-database/40-shutdown-drain-guarantee.md
(chain 3 with 31, status done), board row, slice marker updated
- outstanding and named: a pin below the gate needs new multithreaded test
infrastructure — nothing in runtime/test/ drives wo_engine_start/stop and
no corpus fixture can trigger a stop
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
- T4 monitor + T5 time.after landed in 4092074 (ids 89/90); marker still
listed them pending because it came from master, which lacks that commit
- records the branch baseline (18 suites x 2 flavors, 0 fail) and the gate
at 11 of 12 legs green
- names the stale-artifact trap: after a branch switch, compiler/_build and
runtime/build hold the OTHER branch's binaries, and a v7-vs-v6 mismatch
surfaces only as "no listener"
- the drain guarantee is now the single named blocker; nothing else in the
slice should land before it
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Gate defects, all measured:
- fd check was core-count dependent: `fds_before + 8` read LAZY per-shard
init as a leak. Shards init on first fiber, each taking one io_uring +
one eventfd, capped at nproc; on 20 cores the first wave legitimately
adds 18. Measured 26 -> 44 after 20 clients, still 44 after 40 more.
Replaced with the invariant the check is for: a second wave must not
raise the count. Core-count independent, and catches a slow leak that
any fixed slack would hide
- a failed leg ORPHANED its server: drain inherited $SRV from the soak
leg, so its python died on int("") and the soak server was never
killed — its listener then broke the next run's soak on the same port.
drain now starts its own server; cleanup kills every server a run
started, matched on the run's unique temp dir
- two legs the plan requires were missing: WO_SHARDS=1 (the single-shard
control) and WO_MAILBOX=8 (drop-slow-member backpressure). Both added,
both green. The mailbox leg shrinks the slow client's SO_RCVBUF so it
needs no sleeps
- chat adopted the porch naming (use porch/..., [deps] key) after the
rename landed on master
Decoupling the legs exposed a REAL drain bug, traced and documented in
docs/2026-08-27-chat-drain-finding.md, NOT fixed here:
- on a FRESH server the SIGTERM drain is flaky: 5 of 16 runs left a
client at EOF with no close frame and no diagnostic
- traced: main -> Registry -> Room -> Writer. Registry runs (diag
confirms), the Room NEVER processes its shutdown message, so the
Writer's close branch never runs. Clients that do get a frame are
saved by their own Reader seeing env.stopping()
- ruled out: the spin budget (a 1s wall-clock deadline still failed 2 of
12 — reverted, it fixed nothing and cost 1s per shutdown),
dummy_writer() spawning during shutdown, and write failure
- the fix is an engine guarantee — a send issued before the stop flag is
delivered — which belongs to the actor lifecycle, not a spin count
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Audit found 4 of 10 iterations citing it1 and 4 carrying stale claims the
measurement contradicts.
- 05: framing was contradicted, not merely incomplete. Its goal expected a
gradient to detect ("back-pressure before the cliff"); there is no cliff
— SIGKILL with swap off, exit 0 with swap on, and read latency STEPS
(1us -> 487us) rather than departing. Heading and goal rewritten; the
measurement makes the goal stronger, not weaker
- 05: budget must be bytes — 3.3x footprint spread — with headroom for
index doublings, else it fires during a rehash
- 05: new goal — eviction policy QUALITY is decisive, since getting the
resident set wrong costs 273x, not a few percent
- 06: its revival question now has a reference point. 273x is the KERNEL
SWAP path; `resident: keys` preads via page cache and must beat it. This
file revives only if 5c/5d lands near 273x rather than well below
- 04: write path is not where pressure bites (append ~1%, read 273x), so
the io_uring question that matters is iteration 2's deferred read-path
one, not group-commit
- 00-story: problem statement asserted the store "refuses the insert
rather than dying". Corrected in place — a banner above it was not
enough, a skimmer never reaches it
- residency spec: "swap thrash and the OOM killer" named exits that were
not measured; replaced with silence-or-a-corpse
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Closes the last gap in databasev2 1; gives databasev2 3 its "before".
- `boot` mode: does NOTHING. WO_DATA replay runs before main, so a mode
with no work measures replay plus a fixed startup
- `replayseed N M`: N inserts + M updates — same live rows, longer log
- `replay` leg: empty-store startup floor measured and SUBTRACTED, then
two shapes timed, median of 3 boots each
- premise check: updates must actually append WAL records, else the two
shapes are one measurement and the penalty means nothing
- WAL bytes = non-zero prefix, never file size (fallocate'd to 1 MiB)
- per-record cost stored in NANOseconds: as us it rounded 5.5 and 5.3 to
6 and 5, too coarse for the number a checkpoint exists to improve
- 148 checks, 0 failures; gate bites on a doctored ns_per_record
Measured — same 20 000 live rows, different history:
- 20 000 records: 980 035 B WAL, 110 ms replay, 5.5 us/record
- 40 000 records: 1 960 035 B WAL, 211 ms replay, 5.3 us/record
- 1.9x boot cost for an IDENTICAL dataset; per-record cost flat, so
replay is linear in records not rows
- extrapolated: 10M records ~55 s of boot, 100M ~9 min
- databasev2 3 correction: it planned to use "22's aged-store replay
numbers", which never existed — 22 proved restart correctness, never
timed it
- databasev2 3 hazard recorded: compaction rewrites the log and moves
every record, so it invalidates every `resident: keys` offset — an
arbitrary byte in a rewritten file, not stale-but-readable
- databasev2 1 -> status: done
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
- `randread N R` in the sample: fill N rows, read R across the WHOLE range
- Weyl order `i*2654435761 mod n` — no RNG in the language, none needed;
both legs read the SAME key order so residency is the only variable
- `randread` driver leg: control (256 MiB, does not bind) vs over-cap
(6 MiB + swap), sizes kept modest — quick resolves it in ~5s
- gates the RATIO, not the absolutes: over-cap reads/sec belongs to the
box's swap device, the factor between two runs belongs to the engine
- reads must all resolve (hits == R) or the leg fails; a collapse measured
over unresolved reads is noise
- 133 checks, 0 failures; gate bites on a doctored collapse_x
Measured — this closes the gap the swap leg left:
- resident 1 851 166 reads/sec, p50 0us p99 1us
- over-cap 6 771 reads/sec, p50 128us p99 487us
- 273x throughput, ~480x p99, all 20 000 reads resolving in both
- so the two access patterns sit ~270x apart under identical pressure:
append-mostly insert ~1%, random read 273x
- departure is a STEP not a curve (1us -> 487us, nothing between), which
is why p99_departure_decile finds no knee — there is none
- caveat recorded, NOT inherited: this is demand-paged anonymous memory
through swap (4 KiB/fault, no readahead). `resident: keys` preads via
the page cache — should be better, but databasev2 2 task 7 must measure
its own read path. New criterion added there
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
- `Wide` text-heavy reference shape beside Int-only `Item`
- `growth N int|text`: per-decile RSS read from own /proc/self/status
- `growth-verify`: survivor of a crash must be a contiguous intact prefix
- four footprint legs under a rootless cgroup v2 cap, swap on/off
- `ceiling` leg: die at the cap, then replay must come back intact
- footprint read as median-of-marginals; doublings a separate metric
- 121 checks, 0 failures; footprint gated ±10%, kill-timing ±100%
Measured, and it inverted two of the iteration's own predictions:
- footprint 96.5-100 B/row Int vs 320.6-324 B/row text = 3.3x, NOT the
"order of magnitude" three docs asserted
- table storage has NO checked ceiling: SIGKILL signal 9, not a catchable
WO_T_OOM. overcommit lets malloc succeed; kernel kills on page touch
- swap is NOT latency collapse: 900k rows 148s capped-with-swap vs 150s
uncapped. Append-mostly never re-touches cold pages
- ack-after-fsync survives an OOM kill: ~40k rows, no holes, no corruption
- iteration 2's budget dependency is REMOVED not satisfied — there is no
"swap onset" to derive a fraction from
- fix: subprocess returncode -9 was labelled a "checked refusal"; 137 is
the shell spelling of the same signal
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
- `readiness: ready | refine` is a SECOND axis, orthogonal to status.
`ready` = the brainstorm is complete and the decisions are LOCKED (a spec
approved, or the forks explicitly confirmed). `refine` = open forks remain
and it cannot be planned yet
- `status: refine` RETIRED because it carried both meanings at once, so a held
iteration with an approved spec (language 18, 26) was indistinguishable from
one nobody had thought about. status is now purely where the WORK is:
done | in-progress | pending | hold — `pending` was already the board's own
rendering word, so nothing new was invented
- all 47 iterations classified from EVIDENCE in their own text, not by guess:
"the four forks are SETTLED" / "spec + plan approved" / "Approved spec:" for
ready; "Forks the spec must settle" / "no spec exists yet" for refine. Every
shipped iteration is ready by definition. 19 done, 5 in-progress, 15
pending, 8 hold; 27 ready, 20 refine
- two iterations moved refine -> in-progress rather than -> pending: language
31 and 34 are absorbed into 24 and work on them is literally happening, which
the board already showed as 🔄 while their frontmatter said otherwise. That
disagreement is now gone
- board legend, board-views' frontmatter contract, and two new Dataview
queries updated — the useful one being `readiness: ready AND status:
pending`, the startable set
WHAT THE NEW AXIS IMMEDIATELY SURFACED: of 15 pending iterations, exactly ONE
is startable — databasev2 4, io_uring group-commit, whose forks were confirmed
settled 2026-08-20. Everything else pending needs a brainstorm first. That was
invisible while one key carried both meanings, and it is now on the board.
Also caught by the sweep, unrelated to readiness but found by cross-checking
frontmatter against the board: SIX duplicate rows. Every iteration moved into
databasev2 was still listed in the LANGUAGE pending table under its retired id
(23, 32, 33, 20, 21, 27) as well as its new one. Stale copies removed. And two
databasev2 rows made claims the sweep contradicts — iteration 1 was billed
"startable today" while its forks are open, and 6 still called itself the
ceiling-raiser after 2 took that role.
Docs only. linkcheck 0 broken / 0 anchors.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Task 5c step 1 of docs/superpowers/plans/2026-08-26-table-residency.md, as a
PURE REFACTOR: no storage change, no keys-table anywhere. Borrow is wo_row_ptr
plus a seam, every release is a no-op. Provable on its own before the storage
change it exists to enable.
DESIGN SETTLED BY READING THE STRUCTURES, and both answers make 5c smaller:
- the id hash needs NO new storage. `hvals` is already uint64 holding
slot+1 with 0 = empty (table.h), so offset+1 fits the same field, and the
interpretation is per-table because a table is wholly `all` or wholly
`keys`. No parallel map
- secondary indexes need NO change. `db_ibucket.ids` stores row IDS, not slot
indices, and table.c resolves them through the id hash. I had told the
developer these pointed at slab slots — that was wrong, and it is why this
is one shared accessor rather than 11 rewrites
- the real coupling is the unique shadow: idx_add_row and
row_apply_field_slot both FETCH the conflicting row and compare columns.
Both now borrow/release, so a keys-table's non-resident conflict will be
found rather than silently skipped — a unique check that only examines
resident rows is a correctness hole, not a limitation
The scratch lives on `db_table`, not on the stack and not per call. Per call
would allocate once per candidate inside a bucket loop, turning an O(1) probe
into an allocation storm; a stack buffer is unsafe because the loader bounds
field_cnt at 65535 (loader.c:189), so the worst case is ~512 KB. It is safe
per-table because the store is single-writer, and a `busy` flag is there to
catch a nested borrow rather than let it alias silently. Freed in
table_destroy.
Gates: all 18 runtime suites 0 fail under ASan+UBSan (test_table 856/0,
test_wal 3654/0), oop-e2e 119/0, residency 8/0, employee 8/0, db-actor 8/0.
And the pure-refactor proof the plan asked for: `db-bench --quick` 85/0, every
resident read/query/seed/write floor held — a refactor that moves a number is
not a refactor.
Remaining wo_row_ptr sites for 5d: 6 in table.c, 2 in db.c, 2 in wal.c.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Flagged by the developer: the iteration still described the pre-brainstorm
three-mode design behind a "superseded in part" banner while four tasks had
landed against it.
- iteration 2 rewritten around the shape as built: two keys
(`durable: true|false`, `resident: all|keys`), not a `mode:` enum with
`cold`. status refine -> in-progress
- added a task-by-task progress table with commit hashes, and split the
acceptance criteria into MET (each with how it was verified, not just that
it passed — e.g. the goldens-unchanged claim is `git diff` over golden/
being empty after a WOC_BLESS run, since blessing rewrites all of them) and
OUTSTANDING with the task that owns each
- kept the history rather than deleting it: the three-mode replacement, the
"one real rewrite" that was fiction, and the opposite half that turned out
genuinely deep. An iteration file is where that record belongs
- board row rewritten to agree; the track index's "the lever" section, its
principle-7 paragraph and its sequence rationale all still taught the dead
three-mode design
ITERATION 6 IS NOW LARGELY SUPERSEDED, and bannered as such rather than
quietly gutted. `resident: keys` is the ceiling-raiser and it lives in
iteration 2 (tasks 5c/5d). More than relocated: 6's premise — a user-space
resident working set with faulting and 5's eviction policy — was specifically
REJECTED by the spec in favour of the kernel page cache, since a pread against
a cached page is a memcpy. What may still be left is recorded honestly: revisit
only with a measurement showing the page cache insufficient. Its fork list
survives, especially "does the language surface the fault cost at the use
site", which is still open and still the largest question about what writeonce
is. The sequence rationale is amended too — it had 6 as the ceiling-raiser and
5 as a prerequisite on the critical path; neither holds.
Docs only. linkcheck 0 broken / 0 anchors.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Task 5b of docs/superpowers/plans/2026-08-26-table-residency.md, whose Task 5
is now split 5a-5d (plan updated in this commit).
- the offset twin of wo_row_read (table.c:721): same out-gate contract —
every value handed back is a FRESH VM allocation — but resolved from a file
position instead of the id hash
- fits entirely in wal.c because everything it needs was already public:
scan_record and dec_val are local, and wo_val_decode_vm / wo_db_val_free
are exported at table.h:183-187. Two decode stages, since the record and
the VM speak different dialects: dec_val -> engine slots -> VM copies, with
the engine slots freed as scratch on every path
- ZERO storage change. Nothing calls it yet; that is the point of separating
it from 5c, so the read path can be proven before the slabs are touched
- refuses rather than guessing, each case distinguishable: no intact record
at the offset, a malformed header, a decode failure, trailing bytes, and a
REMOVE tombstone. That last one matters most — handing a tombstone back as
a row would read a deleted row as live
Tested by deep field comparison, not by "it parsed": 24 rows with a nil Text
every third row, each read back BY OFFSET and compared field by field,
including the string bytes. Plus all three refusal paths — tombstone, a
mid-record offset (the silent-wrong-row failure this guards), and past the
intact prefix.
The free-on-every-path claim is VERIFIED, not assumed: removing the free
produced 3 LeakSanitizer reports; restoring it returns to 0. Worth doing
because "ASan is clean" only means something if the harness would have
complained.
PLAN SPLIT: Task 5's storage half was written as if it were plumbing. Measured
instead: wo_row_ptr returns a db_row* into a slab with 11 call sites, table.c
has 37 slab references, db.c:105-181 scans slabs directly, enc_val serialises
FROM the slab, and no operation exists that drops a payload while keeping
index entries. Note this is the OPPOSITE half from the earlier retraction —
the record FORMAT needed nothing, the record STORAGE genuinely is deep. 5c
(id->offset map + drop-payload-keep-index) and 5d (rewiring the call sites,
scans, @unique/FK across the boundary) get their own write-ups.
Gates: test_wal 3654/0 (was 3428), all 18 runtime suites 0 fail under
ASan+UBSan, oop-e2e 119/0, residency 8/0, employee 8/0, db-actor 8/0.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Task 3 of docs/superpowers/plans/2026-08-26-table-residency.md.
- NO LAYOUT CHANGE. The plan said to add descriptor fields; the descriptor
already had a `flags` u32 with only bit0 used, so both properties ride
spare bits (WO_CLASSF_VOLATILE 0x02, WO_CLASSF_RESIDENT_KEYS 0x04). A v7
class record is byte-identical in shape to a v6 one, which is a much
smaller and safer change than the plan assumed
- both spelled as the NON-default, so a zero flags word means exactly what
every pre-v7 image meant: durable, every row resident. A non-@table class
has both clear by construction
- the loader refuses the meaningless pair (bit1+bit2) independently of woc,
on the standing principle that what the loader accepts the interpreter
trusts. Verified by FORGING the flags word in an otherwise valid image,
since woc will not emit one: flags=6 gives "durable:false with
resident:keys", flags=8 still gives "unknown flags"
- WOB_VERSION 6 -> 7. Kept because an OLDER runtime reading a v7 image would
otherwise treat a volatile table as durable and quietly disagree with its
own source. loader.c's check is exact-match, so a v6 image is refused
rather than read with the bits clear — verified by patching a v7 header
back down to 6
GAP FOUND AND CLOSED: woc ACCEPTED `durable: false, resident: keys`. Task 1's
steps covered duplicates and bad values but never the combination, and the
plan had only put that refusal in the loader. The spec wants both, so the
compiler now refuses it too (WO-E102, checked after the argument list is
complete since it is a property of the pair). A compile error is the one a
developer can act on.
VERSION DRIFT: the constant lives in FOUR places, not one. wob.h,
emit.ml:157, disasm.ml:186, and compiler/test/runner.ml:2405 — the last is a
deliberately independent reimplementation of the loader battery, and it
caught the drift as 14 failures rather than silently passing. Its flags mask
and the combination refusal are now in sync too, which is the point of it
being independent rather than shared.
Gates: woc-test 557/0, 18 runtime suites 0 fail, 18 ISO-flavour suites 0
fail, cli_smoke OK, oop-e2e 118/0, employee 8/0, db-actor 8/0, site 21/0.
Zero goldens moved (git diff over golden/ empty).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Task 2 of docs/superpowers/plans/2026-08-26-table-residency.md.
- the check lives in `check_field_types`, which already runs over the raw AST
(so the diagnostic lands at the field's own position, once per declaration)
in Pass 2 with `syms` fully built
- only the durable -> volatile direction is refused. volatile -> durable is
legal: the referencing row is the one that disappears, so nothing is left
holding a stale id
- the message names both classes and both escapes, because "this is wrong" is
less useful than "make Session durable, or declare Order volatile too"
- sees through a `?` wrapper, so `?ref S` is caught too
- code picked as 24 by sweeping `<stage>_prefix ^ "NN"` — 01-23, 25, 26 and
50 were taken, so 24 was a genuine hole. Grepping the literal WO-E224
would have found nothing, which is how ten codes once went missing
- catalogued in the same commit, and the completeness sweep re-run: 53
emitted, 54 catalogued (the extra is retired WO-W201), none missing
PLAN CORRECTION: the plan's second step said to apply the same check to
`backlink` fields. Dropped — a backlink is "NOT a stored column" (ast.ml:72),
so after a restart it resolves to an EMPTY COLLECTION, which is a legal state
indistinguishable from "nothing references me". There is no id to dangle.
Implementing it would have refused correct programs; a spurious diagnostic is
worse than a missing one. A run fixture now pins that the backlink shape stays
legal, so the check cannot silently grow over-broad later.
Gates: woc-test 557/0, oop-e2e 118/0 (was 116 — one compile-fail and one run
fixture added), employee 8/0, db-actor 8/0; employee, db-bench, db-actor,
porch, log-watcher and gc-cycle all still typecheck.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Task 1 of docs/superpowers/plans/2026-08-26-table-residency.md.
- ast.ml: `table_cfg` gains `durable : bool` (default true) and
`resident : residency` (ResAll | ResKeys, default ResAll) — both defaulting
to the pre-existing behaviour, which is what lets every @table written
before this compile byte-identically
- parser.ml: `durable:` takes the existing KwTrue/KwFalse tokens; `resident:`
takes the bare identifiers `all`/`keys`. Given-twice tracked by local seen
flags rather than option fields, so "absent" and "explicitly the default"
stay distinguishable without the AST carrying an option nobody reads
- five new WO-E102 causes, all catalogued in the same commit: durable twice,
resident twice, an unknown resident value, `resident: index` (the
pre-review spelling, with a message naming its replacement), and a retired
design word (mode/store/ram/cold/tiered/paged/mmap/buffer) which gets a
message stating the two real keys instead of a generic "unknown argument"
- dump.ml prints each property ONLY when it differs from its default.
Printing unconditionally would have moved every pre-existing golden, which
this iteration is not allowed to do
- new golden compiler/test/golden/ast/table-residency.wo covers all four
shapes, including a table declaring `resident: all` explicitly and
correctly dumping nothing for it
- verified, not assumed: `git diff --stat` over compiler/test/golden/ is
EMPTY after a WOC_BLESS run, so all 30 pre-existing goldens are untouched.
woc-test 557/0 (was 556), oop-e2e 116/0, employee 8/0; employee, db-bench,
db-actor and porch all still typecheck
- docs: language-surface's @table row now matches what the parser accepts
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
- found during the pre-execution review of the plan, before any code
- the claim was wrong in both spec and plan: table.c's db_val_encode builds
the IN-MEMORY slot; the FILE record is a separate encoding in wal.c and has
been flat since iteration 9. enc_val inlines every kind recursively with no
pointer anywhere; dec_val reads it back; a record is
`WO_WAL_INSERT | class_id | id | <value per field>` in the
len|crc|payload|mark frame; scan_record already preads and CRC-verifies a
record at an arbitrary offset
- so the row encoding needs NO change, and Task 5 (a "self-contained,
offset-based" rewrite billed as the iteration's substantive engineering) is
DELETED, not reduced. 8 tasks -> 7, and the highest-risk task is gone
- the real difficulty is where the spec never looked: wo_wal_append_insert
stages into a 1 MiB buffer, so a record's final offset is unknown until
flush. Threading an accurate offset back through a buffered writer —
correct across partial flush, failed commit and torn tail — is now Task 5's
first two steps, with a unit test that straddles a buffer boundary and a
case asserting no offset is published for a record that never reached disk
- dependent claims corrected: the mmap alternative's premise, the read-path
bullet (now names scan_record/dec_val), and the self-review coverage table,
which records the retraction rather than quietly dropping the row
- root cause worth noting: reading one layer and inferring another. Second
time this iteration — the first was assuming WO_HEAP_MB bounded table
storage when it bounds the VM arena
- no code written yet; linkcheck 0 broken / 0 anchors
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
- 8 tasks, 71 steps, from the 2026-08-26 table-residency spec
- deliberately contains NO code: discarded.md records "raw code in plan
documents" as rejected, and all six preceding plans have zero fences.
Stated in the header so it does not read as an omission. Every step
instead names the exact file and line region plus the required behaviour
- task order is by testable deliverable, not by layer:
1 grammar + defaults (no existing golden may move)
2 the cross-table check — a durable ref into a volatile table is refused
3 .wob v7: descriptor carries both properties, loader refuses the
meaningless combination so it never reaches the engine
4 durable:false skips the WAL at the three existing choke points in db.c;
replay refuses on mismatch rather than resurrecting rows
5 self-contained offset-based records — the one real rewrite, since
table.c returns a malloc'd address as the slot word today
6 resident:keys read path: id->offset map, pread, sequential scan;
@unique and FK-restrict across the boundary are the correctness core
7 the two runtime refusals — durable with no WO_DATA (silent data loss
today), and the resident-footprint budget
8 measure, baseline, crash battery, docs, closeout
- self-review table maps every spec section to a task. Two gaps found and
closed: the escape hatch for an intentionally ephemeral run (a refusal
with no way forward is worse than the loss it replaces), and persisting
the offset map in databasev2 3's snapshot
- linkcheck 0 broken / 0 anchors
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
- `resident: all | keys` replaces `resident: all | index`. Two reasons beyond
taste: it kills the collision with the `index:` argument
(`@table(index: [customer], resident: index)` read badly), and it puts both
values on ONE axis — each now answers "what row data stays resident",
where `all`/`index` mixed a quantity with a structure name
- accurate as well as clearer: what stays resident is the id->offset map, the
secondary indexes and the unique shadows — all key structures; row payloads
are exactly what leaves. `resident: none` was rejected as overclaiming,
since the indexes very much are resident
- checked for collisions: neither `all` nor `keys` is a keyword or a builtin
(`key_at`/`val_at` exist, bare `keys` does not)
- the spec's wart note became a recorded decision; the rejected spelling is
kept quoted so the rationale still reads
- fixes a bug I introduced in the 2026-08-26 track move: all six moved
iterations carried a banner reading "Part of [Story — the database beyond
RAM]" whose link pointed at the LANGUAGE arc — correct target, lying text,
the exact failure mode the link audit warned about. Banners now point at
the databasev2 story, and the original "Part of" line says plainly which
track the iteration was authored in before the move
- linkcheck 0 broken / 0 anchors
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
