Repository navigation
test(ci): cover the three report modes; the loud operator reaches zero (#2492) - #2495
Merged
Merged
Conversation
#2492) Each sat behind a flag CI never passes, and each file says so in its own words: --all is "a report, not a gate", --allow-missing-tools is "reporting nothing, deliberately", --summary prints a counts table. The distinction between a report and a gate is worth keeping. Leaving the report's exit code unmeasured is not part of it: a report that prints its table and then reports failure breaks any script reading it. The opt-out case is the sharpest. check_gate_preconditions proves the ratchet reds when its tools are missing; nothing proved that --allow-missing-tools then returns SUCCESS. An opt-out that fails anyway teaches that the flag does not work, and the next person removes the guard instead of passing it. Loud: zero survivors across thirteen gates. Two report "no success path to break" -- pack_index_consistency_gate and wp18_conformance_gate hold no bare `return 0`, every verdict being a ternary or a SystemExit. They are UNMEASURED in this direction, not covered, and the report says which. Silent: one survivor, the declared one. Its line has moved four times now -- 390, 403, 495, 516 -- and the declaration matched every time, because it names the branch by message. Refs #2492
… none (#2492) check_elab_ratchet offers --allow-missing-tools to accept a missing-tool red locally. A sibling control proved the red; nothing proved the flag then returns success, and the loud operator rewrote that return with nothing noticing. An opt-out that fails anyway teaches that the flag does not work, and the next person removes the guard instead of passing it. Plus: a report is still a program with an exit code -- the report/gate distinction is worth keeping and does not extend to leaving the exit unmeasured. And: "no path to break" is UNMEASURED, not clean. Two gates hold no bare `return 0`; the loud operator skips ternaries on purpose, so those two have no result in that direction and the report must say so. A row that reads as a clean score when it is an absence of measurement is the substitution this skill is about, and it is easiest to commit in your own tool's output. Refs #2492
Contributor
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
Contributor
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
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.
The last three loud survivors were report modes, each behind a flag CI never passes, and each file says so in its own words:
The distinction between a report and a gate is worth keeping. Leaving the report's exit code unmeasured is not part of it — a report that prints its table and then reports failure breaks any script reading it.
The opt-out case is the sharpest
check_gate_preconditionsproves the ratchet reds when its tools are missing. Nothing proved that--allow-missing-toolsthen returns success.An opt-out that fails anyway is worse than none: it teaches that the flag does not work, and the next person removes the guard instead of passing it.
Where both operators now stand
--loudUNCOVEREDwith its reasonTwo rows that are honest rather than clean
pack_index_consistency_gate.pyandwp18_conformance_gate.pyreportno success path to break: they hold no barereturn 0, every verdict being a ternary or aSystemExit. The loud operator takes only bare returns on purpose — forcing a ternary's whole line to 1 is the silent operator seen backwards and would be scored against the wrong control.Those two are unmeasured in that direction, not covered, and the report says which.
The declaration that keeps matching
The one silent survivor's line has moved four times —
:390,:403,:495,:516— under other people's edits and mine, and theUNCOVEREDnote matched every time, because it names the branch by message.Refs #2492