From 8a5c0de16244ddc3082cc3c5fb3cbff4db1e99aa Mon Sep 17 00:00:00 2001 From: Eric Rakestraw Date: Fri, 9 Oct 2026 01:16:14 +0000 Subject: [PATCH] Finalize computational efficiency evidence --- docs/roadmap/implementation.md | 52 ++++++++++++++++++++++++++++++---- docs/roadmap/performance.md | 7 +++-- 2 files changed, 51 insertions(+), 8 deletions(-) diff --git a/docs/roadmap/implementation.md b/docs/roadmap/implementation.md index 507f00d..3ceb03c 100644 --- a/docs/roadmap/implementation.md +++ b/docs/roadmap/implementation.md @@ -3,7 +3,7 @@ Implement the remaining local work in the [performance roadmap](performance.md) in order. Each sequential stage is scoped for one Luna coding-agent prompt. Update status and evidence as each work unit lands. -Stages 1–13 are complete; Stage 14 remains pending. The completed +Stages 1–14 are complete. The completed inspection plan is preserved at `481e2d4:docs/roadmap/implementation.md` in Git history. @@ -545,6 +545,11 @@ redaction, repair, and details-after-completion behavior must continue to pass. ## Stage 14 — Consolidate Evidence And Verify The Complete Local Scope +**Status: Complete (2026-10-09).** The complete local roadmap has been +implemented, ownership decisions and before/after evidence are recorded, and +the final suite, vet, example, formatting, and changed-document link checks +passed. No release or publication work was performed. + **Outcome:** The local roadmap scope is complete with clear measurements and unchanged consumer contracts. @@ -928,7 +933,9 @@ repetitions on the same machine varied substantially; the timing evidence does not establish a reliable latency improvement or regression. The ordinary `Prepare`, `Run`, and public `Details()` allocation counts are unchanged within their recorded baseline ranges; this candidate only changes prepared-execution -creation. +creation. The `B/op` and `allocs/op` values are cumulative allocation per +operation; the separate `retained-B` table is a post-GC live-heap delta for four +handles. They measure different effects and are not inferred from one another. The focused internal snapshot benchmark isolates the same creation path using the Stage 12 fake-collaborator fixture. Times again list five samples, median, @@ -1239,10 +1246,43 @@ attempts continue to use the frozen body. Verification passed: `go test -mod=readonly ./internal/llm ./internal/generation`, and the benchmark family above was rerun without the race detector. +### Final Coverage And Verification + +Stage 13's ownership-transfer implementation, tests, runner documentation, and +decision evidence are committed at `80aecb6` (`Reuse prepared execution +snapshot ownership`). The completed local scope maps to the accepted roadmap as +follows: + +- Item 1's ConfigKit performance request remains recorded at `d53021e`; no + upstream implementation or dependency update was included. +- Items 2–5 are implemented in validation, rendering, request encoding, and + artifact reading. Their before/after fixtures, measurements, and focused + verification appear in the corresponding execution-evidence sections. +- Item 6's ownership review selected the simpler private-payload transfer in + Stage 13. It retains nested-extra and details copies; source isolation, + invalid-copy classification, allocation savings, and fixed-batch heap + observations are recorded above. +- Item 7's lightweight prompt/profile selection and requirement inspection are + implemented and measured in the inspection evidence sections. Production + compiled rendering and assembled offline preparation/execution benchmarks + are also maintained and recorded. + +The canonical runner document describes the new snapshot ownership mechanics. +The source/validation and model-client documents remain consistent with their +implemented flows; no package responsibility, public API, or format semantics +changed. This plan remains the implementation record, and the performance +roadmap's status now reflects completion rather than proposed work. + +Final verification passed on the Stage 13 code: `go test -race -vet=off +-mod=readonly ./...`, `go vet -mod=readonly ./...`, and +`python3 scripts/check-examples.py` (prepare, run, observe). The changed Go +files were `gofmt`-clean, and changed-file Markdown local targets were checked +separately. These checks are local correctness and documentation validation; +they do not add a release gate or imply provider-latency improvement. + ## Open Questions No consumer-policy decisions remain open. File reservation and encoding error -handling have concrete approaches above. The prepared-copy candidate is -deliberately conditional on accepted evidence/ownership criteria, with a -defined retain-current-design outcome; it needs technical verification, not a -new consumer contract or a stronger ownership guarantee. +handling have concrete approaches above. The prepared-copy candidate was +evaluated and retained under the documented ownership and evidence criteria. +No implementation-scope questions remain open. diff --git a/docs/roadmap/performance.md b/docs/roadmap/performance.md index 719c22c..5f4b93c 100644 --- a/docs/roadmap/performance.md +++ b/docs/roadmap/performance.md @@ -1,7 +1,10 @@ # Computational Efficiency Improvements -Status: Accepted scope following review. The ConfigKit request (item 1) is -complete; local performance implementation has not started. +Status: Complete (2026-10-09). The ConfigKit request (item 1) was submitted, +and local items 2–7 were implemented or completed with the evidence and +conditional ownership decision recorded in the +[implementation plan](implementation.md). No dependency upgrade or public +performance guarantee was part of this work. ## Purpose