Repository navigation
tests: CR-150 flat-memory red acceptance tests (l4-pending) - #408
Draft
luisleo526 wants to merge 5 commits into
Draft
luisleo526 wants to merge 5 commits into
luisleo526 wants to merge 5 commits into
Conversation
Add the L0 fixture sources of DESIGN.md section 7: every_k, xsym and scale_out, plus the two F13 fixtures (avg_trade_prefix and selected_open_at_end). The flat-memory acceptance runs build their modules with the release-shaped recipe; no engine code changes. Co-Authored-By: Claude Code <noreply@anthropic.com>
test_report_sink_flat_heap (a'), test_report_sink_metrics (d) and test_report_sink_format (i) are written against the 1.7.0 report sink and fail to compile on this base: pineforge/report_sink.h, src/report_sink_writer.hpp, src/sha256.hpp and src/report_scratch.hpp do not exist yet. That first diagnostic is the recorded fail-before. tests/CMakeLists.txt registers them, and the RSS row, only under PINEFORGE_BUILD_FLAT_MEMORY_TESTS=ON with labels l4-pending;slow;mem, so the current CI build and its ctest floors are unchanged. Co-Authored-By: Claude Code <noreply@anthropic.com>
Acceptance (a), (b), (f), (g), (h), (j) and the INPUTS item 1 borrowed-feed guard. Peak RSS comes from os.wait4 ru_maxrss of the real run_json child, minus every mapped OHLCV byte. Red on the base on the stated slopes; the borrowed guard fails because the export does not exist and run_json clears the feed arrays before the run. Co-Authored-By: Claude Code <noreply@anthropic.com>
Acceptance (d) corpus half and the Python half of (i): sink off vs on over every probe (metrics bit for bit, every E/T value, CSV vs .bin), plus the trailer-vs-hashlib, truncation and -nan self-checks. Exits 1 on this base because strategy_set_report_sink_v1 and --report-sink do not exist. Co-Authored-By: Claude Code <noreply@anthropic.com>
Acceptance (c): 100M synthetic bars with --ohlcv-bin and --report-sink on a 2 GiB spot box, compared against a large-box masked summary and sink sha256. Falls back to the CSV path on this base, where the loader holds about 300 B per bar and the run is OOM-killed. Spot only, not a ctest row. Co-Authored-By: Claude Code <noreply@anthropic.com>
| return | ||
|
|
||
| sink_path = work / "format.sink" | ||
| document = json.loads(run_json( |
| import argparse | ||
| import hashlib | ||
| import json | ||
| import os |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
CR-150 (engine 1.7.0 flat memory), lane L0: the red acceptance tests of the design (section 7, (a)-(k)), written before any implementation.
Each test fails today for the right reason (memory grows, or the sink API does not exist yet). The red runs happen on lab/spot; nothing ran on a Mac. Draft until the red evidence and a gating read are in.
🤖 Generated with Claude Code