Independent dynamic Opus 5 review of PR #980 head25ff4870 found a pre-existing native Reject discrepancy: rejecting pPrChange replaces all of pPr with its CT_PPrBase snapshot, losing live paragraph-mark rPr and sectPr. Comparison AST retains those non-base children correctly.
Reviewer control requires no paragraph merge or mark revision: one paragraph with pPrChange (old jc=both), live rPr/color=FF0000 and sectPr/pgSz. Native rejectChanges yields only jc=both; rejectAllChanges retains the mark color and section. The affected native PhaseF restoration is untouched by PR #980 (current reject_changes.ts around637).
Prioritize section-topology loss above cosmetic empty-rPr cleanup#982. No fix included in PR#980. Word behavior remains unverified; public/synthetic inputs only.
Ref: #941, PR #980 review N1.
Independent dynamic Opus 5 review of PR #980 head25ff4870 found a pre-existing native Reject discrepancy: rejecting pPrChange replaces all of pPr with its CT_PPrBase snapshot, losing live paragraph-mark rPr and sectPr. Comparison AST retains those non-base children correctly.
Reviewer control requires no paragraph merge or mark revision: one paragraph with pPrChange (old jc=both), live rPr/color=FF0000 and sectPr/pgSz. Native rejectChanges yields only jc=both; rejectAllChanges retains the mark color and section. The affected native PhaseF restoration is untouched by PR #980 (current reject_changes.ts around637).
Prioritize section-topology loss above cosmetic empty-rPr cleanup#982. No fix included in PR#980. Word behavior remains unverified; public/synthetic inputs only.
Ref: #941, PR #980 review N1.