geekinney/ekp
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity
888a4019f4
ekp/.phrase/phases/phase-kp-overhaul-20260726/change_narrow_live_append_latency_20260730.md
Kinneyzhang 665164911f perf: make live structural commits incremental 
Reuse prepared paragraph data, safe Elisp DP states, common layout lines, and dirty-island source reconstruction for exact append commits. Simplify C int32 validation and add the frozen task030 evaluator, regressions, and decision records.
2026-07-30 08:24:54 +08:00

2.1 KiB
Raw Blame History

Change: Narrow Live-Append Latency 2026-07-30

2026-07-30 — Implement exact incremental live commits

  • Modify ekp.el: retain prepared paragraph and 1D-DP state in layout plans, retokenize only the last incomplete word plus appended suffix, extend prefix/break/protrusion data from the first dirty box, resume Elisp DP from the earliest state that can reach the new tail, and reuse unchanged layout-line records.
  • Modify ekp-buffer.el: reconstruct only the live dirty island, attempt exact append planning before a full plan, preserve cache identity, and release the old hard-line transaction before the next paragraph becomes live.
  • Modify ekp_c/ekp.c: validate signed 32-bit values with one extract_integer call instead of three Lisp predicate/comparison calls.
  • Add exact append/fallback, multi-engine, unsafe-context, stable-boundary, dirty-island, and invalid-C-vector regressions, plus a frozen baseline/candidate evaluator covering five widths, four paragraph lengths, C/Elisp, and two GC modes.
  • Performance: the four-round source-instrumented evaluator reduces 80-pixel C p95/p99 from 114.717/119.201 ms to 25.490/25.785 ms and Elisp from 588.017/597.093 ms to 43.860/47.578 ms. Ordinary keys remain a zero-work path at 0.595 ms p99 and all widths avoid regression.
  • Production check: with production files byte-compiled, three repeated public-command runs measure append p99 at 1.158–1.326 ms for C and 1.429–1.438 ms for pure Elisp; hard-boundary p99 is 1.251–1.363 ms and 1.457–1.470 ms respectively.
  • Decision: retain the C module. Rust would cross the same Emacs module ABI and cannot remove the Elisp-owned tokenization, measurement, transaction, or publication work. Rewriting the already sub-millisecond C layer would add Cargo and cross-platform release surface without a measured end-to-end gain.
  • Open gate: the locked source-instrumented evaluator still exceeds its absolute 16 ms C/Elisp target, so task030 and issue018 remain open even though the byte-compiled production path is within budget.