test/transcripts/f1b/sabotage-red.txt

1=== F1b sabotage ===========================================

2repo : /home/daviwil/Projects/Code/sigil/task-f1b-folio-doc-reader @ 5b7a5c3e37ed609942921e6e1097fe2112e21dbb

3work : /tmp/tmp.trRFrT3PL9

4============================================================

5 S1 [arm 1, product] RED as predicted -- an unbalanced paren in author.sgl; the build cannot produce the module

6 S2 [arm 2, product] RED as predicted -- the duplicate-id refusal computes its verdict and throws it away

  S3 [arm 2, assertion] RED as predicted -- an expected value in the suite is altered -- the harness notices harness changes

  S4 [arm 3, product] RED as predicted -- the phase refused-witness declaration is deleted; the predicate still refuses it

9 S5 [arm 4, product] RED as predicted -- the 13 disjuncts are added BESIDE the delegating call; behaviour UNCHANGED

  S6 [arm 4, product] RED as predicted -- publication stops descending into a core section -- F1a's reachable defect returns

11 S7 [arm 5, product] RED as predicted -- the folio root line grows one byte; positions stop being comparable

12 S8 [arm 5, product] RED as predicted -- every node's location is synthesized; SHAPE identical, positions moved

13 S9 [arm 6, product] RED as predicted -- the author-form refusal is disabled; the manuscript executes

  S10 [arm 6, product] RED as predicted -- the CONTROL loses a tripwire -- an absence proven by a broken control means nothing

15 S11 [arm 7, product] RED as predicted -- the query marker keeps its prefix and stops naming its query

16 S12 [arm 8, assertion] RED as predicted -- a probe loses its self-test hook -- its green stops meaning anything

17 S13 [arm 9, product] RED as predicted -- the reader imports a subprocess runner

18 S14 [arm 10, product] RED as predicted -- identity collection is neutered; the reader stops refusing duplicates

19 S17 [arm 10, product] RED as predicted -- the duplicate is removed; the inspector must stop reporting one

  S18 [arm 7, product] RED as predicted -- the lowering stops descending into an ADMONITION; folio nodes under one are dropped

  S15 [arm 11, assertion] RED as predicted -- the recorded baseline is altered -- the arithmetic must not be self-satisfying

23COVERAGE BY ARM -- the number that matters. A suite that only ever

24reddens one instrument has not tested the others.

25 arm 1 build <- S1

26 arm 2 reader suite <- S2 (product), S3 (assertion)

27 ...its DERIVED-COUNT clause is DELIBERATELY UNCOVERED. It guards

28 the RUNNER reporting fewer cases than the file declares; the

29 expectation tracks the file BY DESIGN, so no substitution here

30 can separate the two. Leg S16 was written, measured unable to go

31 red, and DELETED rather than kept as false coverage.

32 arm 3 exact membership <- S4

33 arm 4 delegation spelling <- S5 (behaviour-identical), S6

34 arm 5 parity <- S7, S8

35 arm 6 no-execution witness <- S9 (product), S10 (control)

36 arm 7 publication fallback <- S11, S18 (the deeper container descent)

37 arm 8 harness self-tests <- S12 (assertion)

38 arm 9 reader-tier passivity <- S13

39 arm 10 reader/inspector <- S14 (reader half), S17 (inspector half)

40 arm 11 suite arithmetic <- S15 (assertion)

41Every arm has a row. S5 is the load-bearing one: it is BEHAVIOUR-IDENTICAL,

42so it is the only evidence that arm 4's textual half is not decorative.

44F1B_SABOTAGE legs=17 red=17

45F1B_SABOTAGE_COMPLETE all legs red at their named arm, all targets restored by sha256

46SAB_EXIT=0