Repository navigation
Commit a35a64c
CR-150 batch A: L9c + L19o + L19n (integration) (#457)
* cr150 l19lb: executor result for CR-150 lane l19lb (supervisor commit)
Edit-only Codex executor output committed by the supervisor on integration b3c3923. Design, tests, wiring and the compile audit are in the lane's REPORT.md. Not yet proven on the lab; the four-step proof and any parity sweep follow on a squash proof commit.
Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
* cr150 l19lc: executor result for CR-150 lane l19lc (supervisor commit)
Edit-only Codex executor output committed by the supervisor on integration b3c3923. Design, tests, wiring and the compile audit are in the lane's REPORT.md. Not yet proven on the lab; the four-step proof and any parity sweep follow on a squash proof commit.
Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
* cr150 l9c: executor result for CR-150 lane l9c (supervisor commit)
Edit-only Codex executor output committed by the supervisor on integration b3c3923. Design, tests, wiring and the compile audit are in the lane's REPORT.md. Not yet proven on the lab; the four-step proof and any parity sweep follow on a squash proof commit.
Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
* cr150 l19l: executor result for CR-150 lane l19l (supervisor commit)
Edit-only Codex executor output committed by the supervisor on integration b3c3923. Design, tests, wiring and the compile audit are in the lane's REPORT.md. Not yet proven on the lab; the four-step proof and any parity sweep follow on a squash proof commit.
Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
* cr150 l9c: portable report sink (named temp file, per-reader offsets), sink canary line, keep-sink switch (fix round 1)
The temporary report sink is a named mode-0600 file unlinked at close, so every
SinkReader opens its own descriptor on Linux and Darwin (no /proc/self/fd or
/dev/fd branch). run_strategy.py prints
`sink_canary: probe=<id> bytes=<n> records=<n> trades=<n> ok` per probe and
PINEFORGE_KEEP_REPORT_SINK keeps the sink file of the named probes (or all).
Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
* cr150 l19m: exit-phase store on disk, release empty cohort maps and rosters (executor result)
The closed-trade exit phase byte vector (one byte per closed-trade index) moves
to a ScratchFile-backed ExitPhaseStore with one fixed 4 KiB page. The cohort
live_units_by_origin map (104 B per distinct entry id) and the opened roster
vector (24 B per id) are released as soon as they are empty. Written on the old
base; merged onto the L19l head in the next commit.
Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
* cr150 l19n: kernel calendar memo and certified-cycle ring reserve their span buffers once (executor result)
The session-day memo (64 direct-mapped slots) and the certified-cycle ring (4
entries) were already fixed-size; what grew was the heap each slot owned, one
span buffer per new day until the table filled. Every slot's buffer is now
reserved once to a proven per-day bound (session_day_span_bound) and only
assign()ed afterwards, so live heap is constant. Table geometry, slot function
and eviction rule are unchanged. Two new source-free tests compare against the
memo-free calendar.
Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
* cr150 l9c: per-run canary log for full and subset sweeps, canary also checks the E records against the engine receipt (fix round 2)
scripts/run_corpus.sh and scripts/check_corpus_parity.sh keep the sink_canary:/sink_kept: lines of every probe, passing or failing, in one PINEFORGE_CANARY_LOG (unset: no log I/O, behaviour unchanged). New scripts/check_sink_canary_log.py requires exactly one well-formed ok canary per probe of the sweep's own list. run_strategy.py reads the engine's receipt count_e from the live handle and the canary compares the E records a fresh strict read decodes against it; the line gains `equity=<n>` before the verdict. The DESIGN 3.11 wording for the named temporary sink is in the design folder, outside this tree.
Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
* cr150 l19l+l19m: round-2 fixes: exit-phase store and kernel roster store removed, coded throws, portable fault injection, comments restored, citations refreshed
The exit-phase byte vector returns to its base form (its windowing is L8c's); KernelRosterStore is removed (cohort_exists is a range check plus the map); the two uncoded throws in pine_adapter.cpp now carry run-failure codes (report_sink_failed for a corrupt scratch cohort record, engine_invariant for a fixture backend selected after use); the fixture lifecycle is explicit and tested; the Pine prefix of 18 literals in id_pair_store.hpp (kernel-residuals check) is removed; fault injection opens the file by path instead of /dev/fd; the orphan fixture and a stale process comment are removed; the four deleted comment sentences of the base are restored; two witnesses (live settlement capacity release, part C typed failure) are added; 426 doc citations across seven documents are refreshed (digits only).
Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
* docs: re-anchor file:line citations after merging integration 5022017
Digits only, no prose changed: every citation into a file this lane or L8c changed was recomputed from the final tree.
Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
* cr150 l19l: reversal cancels exits of every known cohort id again, not only the resident ones (behaviour fix, witness test)
L19l-A's disk-resident cohort table made the reversal cancellation loop (on_applied, full close) visit resident rows only, while a spilled cohort with no exposure can still own queued exits; cancellation is an action over the id, so spilling an eligible row suppressed it and changed command events, pending requests, hashes and later trades. The loop now snapshots the same two exemptions as the base (the applied snapshot's source id and any live entry opening on the new side) and streams every logical id in lexicographic order through the new DiskResidentIdTable::for_each_sorted_id, without restoring facts or building a history-sized vector. tests/test_adapter_reversal_storage.cpp runs a live reversal fixture with capacity release and cohort eviction on and off against the retained-storage run, comparing every trade field, recorded and callback hashes and the final hash.
Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
---------
Co-authored-by: Claude Sonnet 5.5 <[email protected]>1 parent 5022017 commit a35a64c
47 files changed
Lines changed: 6420 additions & 436 deletions
File tree
- docker
- docs
- adr
- design
- pages
- include/pineforge
- detail
- source
- scripts
- src
- source
- tests
- cr150.d
Some content is hidden
Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
5282 | 5282 | | |
5283 | 5283 | | |
5284 | 5284 | | |
5285 | | - | |
5286 | | - | |
5287 | | - | |
5288 | | - | |
5289 | | - | |
5290 | | - | |
| 5285 | + | |
| 5286 | + | |
| 5287 | + | |
| 5288 | + | |
| 5289 | + | |
| 5290 | + | |
| 5291 | + | |
| 5292 | + | |
| 5293 | + | |
| 5294 | + | |
| 5295 | + | |
5291 | 5296 | | |
5292 | 5297 | | |
5293 | 5298 | | |
5294 | 5299 | | |
5295 | 5300 | | |
| 5301 | + | |
5296 | 5302 | | |
5297 | | - | |
| 5303 | + | |
| 5304 | + | |
5298 | 5305 | | |
5299 | 5306 | | |
| 5307 | + | |
5300 | 5308 | | |
5301 | | - | |
5302 | | - | |
| 5309 | + | |
5303 | 5310 | | |
5304 | 5311 | | |
5305 | 5312 | | |
5306 | 5313 | | |
5307 | 5314 | | |
5308 | | - | |
| 5315 | + | |
5309 | 5316 | | |
5310 | 5317 | | |
5311 | 5318 | | |
| |||
5340 | 5347 | | |
5341 | 5348 | | |
5342 | 5349 | | |
5343 | | - | |
| 5350 | + | |
| 5351 | + | |
| 5352 | + | |
| 5353 | + | |
| 5354 | + | |
| 5355 | + | |
| 5356 | + | |
5344 | 5357 | | |
5345 | 5358 | | |
5346 | 5359 | | |
| |||
Large diffs are not rendered by default.
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
218 | 218 | | |
219 | 219 | | |
220 | 220 | | |
221 | | - | |
| 221 | + | |
222 | 222 | | |
223 | 223 | | |
224 | 224 | | |
| |||
Large diffs are not rendered by default.
Large diffs are not rendered by default.
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
25 | 25 | | |
26 | 26 | | |
27 | 27 | | |
28 | | - | |
| 28 | + | |
29 | 29 | | |
30 | 30 | | |
31 | 31 | | |
| |||
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
502 | 502 | | |
503 | 503 | | |
504 | 504 | | |
505 | | - | |
| 505 | + | |
506 | 506 | | |
507 | 507 | | |
508 | 508 | | |
| |||
1496 | 1496 | | |
1497 | 1497 | | |
1498 | 1498 | | |
1499 | | - | |
| 1499 | + | |
1500 | 1500 | | |
1501 | 1501 | | |
1502 | 1502 | | |
| |||
2047 | 2047 | | |
2048 | 2048 | | |
2049 | 2049 | | |
2050 | | - | |
| 2050 | + | |
2051 | 2051 | | |
2052 | 2052 | | |
2053 | 2053 | | |
2054 | | - | |
| 2054 | + | |
2055 | 2055 | | |
2056 | 2056 | | |
2057 | 2057 | | |
| |||
3696 | 3696 | | |
3697 | 3697 | | |
3698 | 3698 | | |
3699 | | - | |
| 3699 | + | |
3700 | 3700 | | |
3701 | 3701 | | |
3702 | 3702 | | |
| |||
0 commit comments