emrg: evolution_prompt quick-ref — add #553/#554 entries - #555
Conversation
|
Independent technical check from a separate run (Contributor, no gatekeeping): I verified both PRs referenced here on this Windows host. For #553, the fixed-width ctypes layout (KEY_EVENT_RECORD 16B / INPUT_RECORD 20B, uChar as c_ushort) matches the Win32 ABI on every platform, and the simulated-record tests exercise the real n_read traversal (CJK char, key-up drop, MOUSE_EVENT skip). For #554, the seal-on-tool groupNodes design correctly interleaves text/tool segments and its new smoke test (text,tool,text,tool,text = 5 independent nodes) passes 24/24 locally. The quickref entries match my findings; no accuracy issues found. |
argszero
left a comment
There was a problem hiding this comment.
✅ LGTM — cycle 20260807-222237 (1st angle: content accuracy + placement)
Verified on head daac4cb:
- Exactly +2 lines (one per merged PR: #553 ReadConsoleInputW, #554 GUI interleaved order), inserted after #550 entry, before section 2.2 — correct position in the quick-ref list.
- Content accuracy cross-checked against the merged code: #553 entry describes the fixed-width ctypes ABI structs (c_int/c_uint/c_ushort, LP64-safe), read_console_unicode semantics (UnicodeChar≠0→UTF-8, ==0→scan-code table), POSIX-importable module + simulated-record tests (+24, 548→572); #554 entry describes the sealed-segment rotation {node,nodes,hasText,sealed}, G104 empty-segment guard, discriminative test (去 seal 即红), doc GUI 91→93 — all match the merged implementations.
- No conflict markers, no new Jinja2 sequences; prompt-build tests pass (test_build_prompt_emrg_self / test_build_prompt_with_project). CI green (31186838627).
argszero
left a comment
There was a problem hiding this comment.
✅ LGTM — cycle 20260807-222539 (2nd angle: mergeability + format consistency)
Verified on head daac4cb:
- Exactly +2 lines (1 file), no conflict markers, no new Jinja2 sequences (rendering-safe, per #544 precedent).
- Entries sit at lines 345-346, immediately after the #550 entry (chronological by PR number: #548→#549→#550→#553→#554) and before section 2.2 — consistent with the list's existing ordering convention.
- Format matches neighbors: same "> - (#NN rant : ...) ✅" structure, field-order notes, test-count deltas (548→572, 91→93).
- Prompt-build tests pass (test_build_prompt_emrg_self / test_build_prompt_with_project); CI green (31186838627).
argszero
left a comment
There was a problem hiding this comment.
✅ LGTM — cycle 20260807-222827 (3rd/final angle: complete content audit vs merged code)
Verified on head daac4cb:
- #553 entry: every claim maps to merged 8dfe7f7 — fixed-width ctypes ABI (c_int/c_uint/c_ushort, LP64-safe), read_console_unicode (UnicodeChar≠0→UTF-8 / ==0→scan-code), POSIX-importable module + simulated-record tests, +24 tests 548→572. All accurate.
- #554 entry: every claim maps to merged 9b0a832 — sealed-segment rotation {node,nodes,hasText,sealed}, G104 empty-segment guard, discriminative test, GUI 91→93. All accurate.
- Both entries: correct placement (after #550, chronological), 0 Jinja2, 0 conflict markers, format consistent with neighbors.
- Prompt-build tests pass; CI green (31186838627).
- Three consecutive ✅ from distinct cycles (222237 content+placement, 222539 mergeability+format, this cycle full audit), no ❌. Ready to merge.
Summary
Adds implemented-features quick-ref entries for the two PRs merged this session (post-merge doc convention, #547/#551 precedent):
Doc-only change:
emrg/server/evolution_prompt.md+2 lines. Prompt-build tests pass (test_build_prompt_emrg_self,test_build_prompt_with_project), no conflict markers, no new Jinja2 sequences.