E.4.PFIP:5 - Archetypal Grounding

Tell — one-to-one text revision. A DPF maintainer adds an accepted pattern and updates one existing pattern. The source-to-candidate comparison confirms both changes. A text comparison with the predecessor monolith then finds an unrelated non-use boundary deleted during paragraph cleanup. That boundary has no accepted retirement decision, so the preservation conclusion reports accidental loss even though the build and accepted-source comparison are clean.

Show — one form split into two. One predecessor pattern-text form is split into two candidate forms. The accepted split decision names all three expressions. The complete predecessor inventory allocates a shared working-situation claim to both candidates, three predecessor claims to one candidate pattern card, and one obsolete tool instruction to an accepted content-retirement decision. A predecessor non-use warning appears in neither candidate. Retiring the old form does not retire the warning, so the missing line blocks preservation until the warning is restored or its content change or retirement is accepted.

Show — architecture diagram. One FPF architecture-diagram form is replaced one-to-one by a candidate diagram for the same orientation use. The diagrams have no useful line diff. The maintainer uses the FPF pattern that defines or constrains that diagram form to select named framework parts, dependency edges, and legend distinctions for the inventory. The maintainer uses C.33 to record captured and lost structure and C.34 to record the declared correspondence. One predecessor dependency edge has no candidate correspondent, so similar layout cannot support a positive conclusion.

Show — local practice framework. An LPF candidate replaces a repeated general explanation with a reference to a new FPF method. The accepted-source comparison passes. The predecessor comparison still checks the LPF’s local recovery cue, concrete work-size limit, and tool-specific stop because the general FPF method does not include those local actions. Removing them would be loss, not successful deduplication.