issues dated: the rule was disclosed in one command and hidden in its sibling, in the same file - #3198
Merged
gHashTag merged 1 commit intoSep 4, 2026
Conversation
Third confirmed finding from the #3195 fan-out. `tri issues dated` printed "no figure in the title 205". The bucket comes from `carries`, whose digit scan accepts a run only at `i - start >= 2` -- two or more digits, boundary-clean -- so #2627 "Census the optimizer: 4 of 7 passes have no precondition" carries no figure. Measured on gHashTag/t27 over the same 509-issue read: 21 of the 205 carry a single-digit figure. The label promised the reader 184. The disclosure already existed forty lines away. `tri issues numbers` prints "single-digit only, excluded 21 (--single prints them)" from the identical rule in the identical file. One command names the exclusion its rule makes; its sibling never mentioned it. A caveat that exists in one command is not a caveat the reader of another one ever sees. The count stays -- it is the complement of the population, so the three lines still add up -- and the label names the rule: no figure the TWO-digit rule reads 205 of which carry a SINGLE digit 21 (`tri issues numbers --single` prints them) The second line is absent, not printed as a zero, when nothing was excluded: a caveat that prints when there is nothing to caveat teaches the reader to skip the line. Third pass running whose surviving mutant was the WIRING rather than the function: last_pass in red.rs, claims_seen in gates.rs, and single_digit_only here. In all three the function was correct and well tested and one line elsewhere put its result to no use. Killed by a structural test reading the call site, needle split across two literals. SKILL 545. Refs #3195
This was referenced Sep 4, 2026
Contributor
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
Contributor
gHashTag
deleted the
loop/disclosed-in-one-command-hidden-in-its-sibling
branch
September 4, 2026 20:27
gHashTag
added a commit
that referenced
this pull request
Sep 4, 2026
#3198 squash-merged, so this branch's copy of that commit conflicts with master's. Only SKILL 546 is new here; 543-545 are identical on both sides. Resolved BY TITLE with the same guard as before: every one of master's section titles must survive the merge, numbers must be unique and ascending, and no NEW conflict markers may appear -- anchored to the start of a line, because this file DOCUMENTS conflict markers in prose. Refs #3195
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.
Third confirmed finding from the #3195 fan-out.
tri issues datedprinted:The bucket comes from
carries, whose digit scan accepts a run only atif i - start >= 2— twoor more digits, boundary-clean. So
#2627 "Census the optimizer: 4 of 7 passes have no precondition"carries no figure, because4and7are one digit each.Measured on
gHashTag/t27over the same 509-issue read: 21 of the 205 carry a single-digitfigure. The label promised the reader 184.
The disclosure already existed, forty lines away
tri issues numbers, from the identical rule in the identical file, prints:One command names the exclusion its rule makes; its sibling, over the same rule, never mentioned it.
A caveat that exists in one command is not a caveat the reader of another one ever sees.
After
The count stays — it is the complement of the population, so
509 = 205 + 304still closes — and thelabel names the rule that produced it. The second line is absent, not printed as a zero, when
nothing was excluded: a caveat that prints when there is nothing to caveat teaches the reader to skip
the line.
Three passes, three surviving mutants, all of them the wiring
last_passinred.rsclaims_by_scopeingates.rsclaims_seen +=adding both halves at the call siterender_no_figureinissues.rssingle_digit_onlynever called, so the tally is always 0In all three the function was correct and thoroughly tested, and one line elsewhere put its result to
no use or the wrong use. Nothing but mutation found any of them. Each needed a structural test
reading the call site, with the needle split across two literals — because the first such test in
this series searched the file for a string it also contained and passed against its own mutant.
4 mutants run here, 3 killed immediately, the 4th after the structural test. 711 crate tests.
SKILL 545. Refs #3195