test(pixelflow-codegen): close x86_64.rs mutation gaps by jppittman · Pull Request #1054 · jppittman/pixelflow

jppittman · 2026-08-31T11:47:04Z

Summary

Scheduled continuation of the test-quality-audit series
(docs/bugs/*-test-quality-audit-followup.md). This pass picks up
pixelflow-codegen/src/emit/x86_64.rs — the x86-64 SSE/VEX raw
instruction encoder — flagged since the 2026-08-08 audit as never
mutation-tested.

While this branch was in flight, main merged eight commits (#1055–#1062)
that rewrote the whole emit module around a shared Vex/Digit builder
abstraction and moved the collapse-loop driver logic into the per-ISA
files. x86_64.rs grew from 864 to 1647 lines. The first draft's ~35
tests targeted functions that refactor deleted or reshaped, so this PR
merged main in and started the mutation pass over against the current
file. Full arc documented in docs/bugs/2026-08-31-test-quality-audit-followup.md.

Removed emit_xorps/emit_andps: pub fns with zero callers anywhere
in the repository. Neg/Abs already go through the VEX
emit_vxorps/emit_vandps instead.
Widened the mutants test filter to include tests/collapse_loop.rs
(an existing 11-test integration suite that compiles and executes
collapse-loop kernels), which turned out to already cover a whole
21-mutant category in X86Backend's driver/scaffold plumbing —
invisible only because this series has always scoped cargo mutants
to --lib. No unit tests were needed there; the audit doc adds a
methodology note for future passes to check this before writing new
tests against a mutants report.
Added 15 new tests for what remained: genuine logic bugs kept at this
file's real API boundary (byte-in/byte-out for the encoder primitives,
plus two frame_bytes > 0 boundary conditions in X86Backend::prologue/
epilogue that are invisible to any value-comparison test since the
wrong branch only emits dead-but-harmless bytes).
cargo mutants -p pixelflow-codegen --file .../x86_64.rs -- --lib --test collapse_loop: 388 mutants, 334 caught, 39 documented equivalent
(replace | with ^ on bit-disjoint ModRM/VEX/REX fields — provably
identical for every input), 0 real gaps.

Test plan

cargo test -p pixelflow-codegen (all targets incl. doctests and
collapse_loop.rs): pass
cargo build --workspace: clean
cargo clippy -p pixelflow-codegen --lib --tests: clean
cargo fmt -p pixelflow-codegen -- --check: clean
cargo mutants -p pixelflow-codegen --file pixelflow-codegen/src/emit/x86_64.rs -- --lib --test collapse_loop: 388 mutants, 334 caught, 39 documented equivalent,
0 real gaps

🤖 Generated with Claude Code

https://claude.ai/code/session_01BR5xrsVNGXHBceptcxBvui

Adds byte-exact unit tests for the x86-64 SSE/VEX instruction emitters (movaps load/store, f32 constant embedding, the general VEX encoder, context-pointer load, vpextrd/vmovss gather helpers, rsp-relative spill store/load, ternary MulAdd, epilogue, movmskps, cmp eax) that previously had zero direct coverage beyond end-to-end float-value comparisons. Removes emit_xorps/emit_andps: dead code with zero callers anywhere in the repo (Neg/Abs use the VEX emit_vxorps/emit_vandps instead). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BR5xrsVNGXHBceptcxBvui

The offset=128 boundary tests for emit_movaps_load/store used Reg(0), so the ModRM reg-field shift (<< vs >>) was a no-op either way and the mutant survived. Switch to a nonzero register to distinguish them. Also cargo fmt. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BR5xrsVNGXHBceptcxBvui

…aps) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BR5xrsVNGXHBceptcxBvui

github-actions · 2026-08-31T11:47:15Z

🤖 Hi @jppittman, I've received your request, and I'm working on it now! You can track my progress in the logs for more details.

… against upstream Vex-builder refactor Upstream (#1055-#1062) replaced the raw byte-pushing SSE/VEX helpers with a Vex/Digit builder abstraction and merged regalloc.rs's driver logic into this file. Took main's version of x86_64.rs wholesale; the byte-exact tests added in this branch targeted functions that no longer exist and need to be rewritten against the current API in a follow-up commit. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BR5xrsVNGXHBceptcxBvui

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit:e077cb8118

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

… Vex-builder refactor Rewrites the mutation-gap-closing tests against upstream's new Vex/Digit builder API (main moved to it while this branch was in flight, via #1055-#1062). Kept the scope to this file's real API boundary and to genuine logic bugs rather than pinning private-method delegation: - Vex::head's W-bit shift (real bug: `<<` vs `>>` only diverge when w=1, which no current caller passes, so the encoder is tested directly rather than through a production call site). - emit_movups_store_base's REX.R/B computation (four register high/low combinations). - emit_load_ptr_from_ctx, emit_movmskps_eax, emit_cmp_eax_imm8 — this file's byte-in/byte-out contract. - x86_redzone_disp's negation and disp8 overflow boundary. - emit_binary_safe's aliasing decision (the one case that must route away from the plain two-operand `emit_binary` call). - X86Backend::prologue/epilogue's frame_bytes>0 boundary: a real gap invisible to any execution/value test, since the wrong branch only emits a harmless dead `sub/add rsp, 0` pair, not a wrong answer. Most of the remaining `cargo mutants` misses in this file are `| to ^` inside ModRM/VEX/REX byte construction — provably equivalent, since those operands are always bit-disjoint fields by construction. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BR5xrsVNGXHBceptcxBvui

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit:4435869ff6

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

…factor Documents the full arc: the first draft's tests targeted x86_64.rs before upstream's #1055-#1062 rewrote it around a Vex-builder abstraction, the merge that discarded the stale draft, and the finding that a whole class of apparent driver/scaffold gaps was actually already covered by tests/collapse_loop.rs -- invisible only because this series' mutants filter has always scoped to `--lib`. Adds a methodology note recommending future passes check for a relevant integration-test target before writing new unit tests against a mutants report. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BR5xrsVNGXHBceptcxBvui

…ommit)

Records the AVX2/AVX-512 backend mutation-testing pass: methodology notes on cargo-mutants' cfg-blindness (a "missed" mutant on this file can be dead code under the active RUSTFLAGS rather than a real gap), the confirmed-zero real-gap final state for both tiers, and an updated backlog noting the four unmerged draft PRs (#1049-#1051, #1054) from prior passes that this session did not touch.

chatgpt-codex-connector · 2026-09-01T21:15:47Z

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.

…-ABI refactor main's #1082 ("one kernel ABI, one compile entry") landed while this branch was up to date, deleting IsaBackend::prologue/epilogue (and the `if self.frame_bytes > 0` conditional they gated) in favor of frame_alloc/frame_free, and folding emit_movups_store_base into emit_movups_store<D: Disp>. The merge from main therefore left this branch's own tests referencing functions that no longer exist. - emit_movups_store_base_* tests now drive emit_movups_store through a NoDisp address, which is the same encoding the deleted function produced. - x86_redzone_disp's overflow-boundary test compares against CompileError::Internal(..) instead of a bare &str, matching #1081's CompileError migration. - The four prologue_*/epilogue_* tests are removed rather than rewritten: the frame_bytes>0 conditional they exercised was deleted outright by #1082, not renamed or moved. frame_alloc/frame_free are unconditional delegations to emit_sub_rsp/emit_add_rsp now, called by the shared collapse-loop scaffold with a total that is never zero, so there is no surviving branch for these tests to target. cargo test -p pixelflow-codegen (all targets incl. collapse_loop.rs): pass. cargo clippy -p pixelflow-codegen --all-targets: clean. cargo fmt -p pixelflow-codegen -- --check: clean. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NV5ZT2mxTnBC66unvBE7FF

… segregate / delete inventory (#1090) ## Summary One docs file, `docs/plans/2026-09-01-dead-code-with-ideas.md`, answering JP's 2026-09-01 prompt ("I think we still have a lot of dead code with some decent ideas in it. factored.rs and so on"). Read-only audit of `origin/main` @ `83015dcd` plus the open branches for #1044, #1083, #1084, #1085, #1072, #1086. **No source changes** — JP decides deletions. Every row cites `file:line` and the caller grep that produced its status. Verdicts follow the segregate-don't-delete doctrine plus one new verdict for this pass, REUSE-R2 (a dormant piece whose idea the Guide's candidate-context design needs, with the plug-in point named). ## Verdict counts | verdict | items | LOC | |---|---|---| | KEEP (live) | 18 | ~16,900 (all live callers verified) | | REUSE-R2 | 6 | ~3,900 file / ~2,150 non-test — `GraphAccumulator` binding primitive, `unified_backward` adjoint, `episodes.rs` budget sampler, `init_with_latency_prior`, `bilinear_score`/`encode_rule_from_arena`, `pixelflow-ml` SH/RFF family (idea only) | | SEGREGATE | 5 | ~2,180 (4 stale docs → archive; whole-graph scoring tower already segregated; #994 is JP's) | | DELETE | 6 | **~1,357 Rust LOC on main** + 367 lines of stale doc + close#1044 (2,690 LOC that never lands) | ## Things the scans did not expect - `CostModel::load_or_default` (`cost.rs:410`) has **zero callers** and probes `$HOME/.config/pixelflow/cost_model.toml` and a nonexistent `pixelflow-ml/data/`; its only writer is `pixelflow-core/src/bin/calibrate_costs.rs` (Jun 11, superseded by `measure_latency_prior.rs`). ≈700 LOC incl. the tests #1051 merged today — decision D3. - The four `pixelflow-pipeline/examples/{op_cost_stress,egraph_choices,ilp_benchmark,critical_path_test}.rs` touch no pixelflow crate at all (scalar `f32` loops) and carry `cargo run -p pixelflow-ml` headers — 636 LOC, DELETE. - The 2026-09-01 candidate-context design branch (`claude/phase3-context`) is not on origin; the on-main design of record is `2026-08-31-guide-design-revision.md` §4, which already argues candidate-local over whole-graph. #1084 built `CandidateFeatures` without reusing `GraphAccumulator` — the reuse is the `shift_by`/Hadamard binding primitive at candidate scope, not the accumulator struct (seam R1). ## Open-PR disposition (§5) #1044 close + cherry-pick its 2 journal lines; #1049 close/fold; #1051 merged; #1054 merge after #1086's one caveat; #1050 already closed (moot as predicted); #994 JP's. Consistent with #1086's landing order (#1053 → #1085 → #1084/#1083). ## Test plan Docs only. No `cargo` gates affected. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: JP Pittman <jppittman@jpptech.dev> Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

…cted (#1089) ## Summary Second pass over the open PRs, continuing #1086. That PR did the rebase sweep and wrote recommendations; this one records what was **executed** against them and re-derives the board rows that have since gone stale. Adds one document: `docs/results/2026-09-01-open-pr-sweep-followup.md`. ## What was executed | Action | PR | Basis | |---|---|---| | Merged | #1051 — `cost.rs` mutation gaps | test + docs only, 0 unresolved threads, all blocking checks green | | Merged | #1049 — `graph.rs` test renames | test + docs only, 0 unresolved threads, all blocking checks green | | Closed | #1050 — `regalloc.rs` mutation gaps | target code verified absent from `main` and from its own branch | | Reviewed | #1044 | `shepherd` label; found it had never been reviewed | | Retitled | #1087 | `CL metadata` wanted a conventional-commit prefix | After merging, all 11 remaining open PRs were re-checked with `git merge-tree --write-tree` against the new `main` — **all still merge cleanly**, so the merges introduced no conflicts. ## Corrections to #1086's board #1086 was accurate when written and is now one step behind. The one that moved against the trend: **#1054 went from green to red.** #1082 removed `X86Backend::prologue`/`::epilogue` and #1081 changed the error type, so its tests no longer compile (`E0599` ×4, `E0308`); Clippy, both test jobs and the ISA matrix are red. This is the *second* encoder refactor to invalidate the branch — the first is already in its own history as `4435869f`. Recommendation is to hold it until that file stops moving rather than fund a third re-close pass. ## Three things #1086 could not have seen - **The saturation collision has five participants, not three.** #1044's `variants.rs` calls `eg.saturate_with_limit(64)` (lines 229, 262), which #1085 deletes. Disjoint files, so git merges clean and the *build* breaks on whichever lands second. - **#1083 and #1087 are two mechanisms for one fact** — a real `SaturationStopReason` field vs. outside-in inference — on confusingly adjacent branch names. #1083 went green during the sweep. One should be picked before either lands. - **#1044's zero unresolved threads is an artifact.** Codex hit its usage limit on 2026-08-28 before reviewing it, so thread count ranks an unreviewed 3060-line diff as the cleanest thing in the set. ## Verified rather than relayed #1072's call-overhead P1 is real: `bench_extraction_3way.rs:2607` aggregates `bench.ns * normalization` while `adjusted_ns` is only serialized at `:2591`. Worth noting the direction — adding a constant to both arms pulls the ratio toward 1, so the true regression is *larger* than the reported 1.0153. The qualitative verdict survives; the intervals do not. ## What this pass could not do The remaining gap to "no unresolved comments, no CI failures" is entirely commits belonging on other branches, which this session is not scoped to push. The four outstanding fixes (#1083 `saturation-telemetry = ["std"]`; #1053 stale script header; #1044 `saturate_with_limits`; #1087 `cargo fmt`) are written down in the doc with exact locations. Roughly 60 unresolved threads remain across 8 PRs, 13 of them P1, all filed by `chatgpt-codex-connector` — no human review is unaddressed. ## Test plan Documentation only — no code changes, so `cargo` gates are unaffected. Branch state claims were produced from `git merge-tree --write-tree` and `git rev-list --count` per branch; CI and review-thread claims from the GitHub check-run and review-thread APIs; the `#1072` and `#1044` source claims by reading the files at those branches' heads. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01BAj1WWiTMvAJ6qK42LNyD6 --- _Generated by [Claude Code](https://claude.ai/code/session_01BAj1WWiTMvAJ6qK42LNyD6)_ Co-authored-by: Claude <noreply@anthropic.com>

Rewrites the 2026-09-01 sweep for the state at 2026-09-02 10:10 UTC, main at 44c9fa3 — seventeen commits later. #1089 covers the intervening window and is not repeated. The collision the first pass flagged has landed and taken the board with it. #1083, #1085, #1107 and #1108 all touched one seam and merged overnight; ten of the fifteen open PRs now conflict with main, where yesterday all thirteen merged cleanly. Nine of the ten conflict in some subset of egraph/graph.rs, egraph/saturate.rs, egraph/mod.rs and runtime.rs. The cheap mitigation was to rebase the queue after the first of the four landed, not after all four. Also records what became of the first pass's twelve recommendations — nine resolved, either merged or closed — and two corrections to it: - Thread count is not a quality signal. #1044 was ranked merge-ready partly on "zero unresolved threads", which was an artifact of the review bot hitting its usage limit before ever reading a 3,060-line diff. - The saturation collision had five participants, not three; #1044's variants.rs called a method #1085 deletes, in a disjoint file. A clean merge-tree is not evidence that a branch still builds. Closure recommendations for the three PRs that did not move: #1054 (red against main — X86Backend::epilogue removed by #1082, CompileError by #1081 — and invalidated by a second encoder refactor), #1072 (modify/delete against bootstrap_extraction_head.rs, which #1093 deleted, so it documents a harness main has removed), and a decide-either-way on #994. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NV5ZT2mxTnBC66unvBE7FF

chatgpt-codex-connector · 2026-09-02T10:19:28Z

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.

jppittman · 2026-09-02T10:19:49Z

Brought up to date with main (44c9fa3f) and fixed — this was red and is now building.

What broke it. The branch was green at 136cc63 and went red the moment it was brought up to date, with no textual conflict anywhere: git merged it cleanly and it did not compile. Nine errors in pixelflow-codegen/src/emit/x86_64.rs — X86Backend::epilogue no longer exists (#1082's one-kernel-ABI refactor deleted prologue/epilogue in favour of frame_alloc/frame_free), plus E0308 mismatches from #1081's &'static str → CompileError migration. It failed on ubuntu, macOS, Clippy and all three ISA levels.

The fix (a42a16d5) reconciles this branch's own tests with that refactor:

emit_movups_store_base_* now drive emit_movups_store through a NoDisp address — the same encoding the deleted function produced.
x86_redzone_disp's overflow-boundary test compares against CompileError::Internal(..) rather than a bare &str.
The four prologue_*/epilogue_* tests are removed rather than rewritten, and that is worth justifying rather than glossing: the frame_bytes > 0 conditional they exercised was deleted outright, not renamed or moved. I verified prologue/epilogue appear nowhere in the workspace, and that frame_alloc/frame_free are unconditional one-line delegations to emit_sub_rsp/emit_add_rsp (x86_64.rs:1298-1304), called by the shared collapse-loop scaffold with a total that is never zero. There is no surviving branch for those tests to target, so this is deleting tests for deleted code, not dropping tests to get green.

Gates run against current main, actual output:cargo test -p pixelflow-codegen --lib 125 passed / 0 failed; cargo test --workspace all suites passing; cargo clippy --workspace --all-targets -- -D warnings clean; cargo fmt --all --check clean.

One thing still open, and a recommendation. This is now the second encoder refactor to invalidate this branch — the first, #1055–#1062's Vex-builder rewrite, is already in its own history at 4435869f, which discarded and rewrote the original tests. The audit doc's "0 real gaps" conclusion has never been re-verified since either refactor, and emit_vpextrd_to_gpr / emit_vmovss_load_scaled in particular carry no coverage claim in the doc at all. Before merging, either rerun cargo mutants -p pixelflow-codegen --file pixelflow-codegen/src/emit/x86_64.rs -- --lib --test collapse_loop and record the real numbers, or soften that conclusion to state what was and wasn't verified. Given how fast this file is moving, re-running the mutation audit once it settles may be the better trade than a third re-close pass.

Generated by Claude Code

fixed The remaining conflicted branches are not one undifferentiated pile. Each is two separable jobs, and conflating them is what made the set look intractable: (a) The core delta is superseded and mechanical. Every branch carries its own answer to "why did saturation stop", written before #1083 landed one. #1087 is the clearest: ten graph.rs hunks of `bool truncated` against main's `ScanStop { Completed, ClassCap, Deadline }` — the same fact at strictly more resolution, and this codebase's own "extend the type, not the convention" rule already applied on the main side. Nothing to weigh; take main's. (b) The harness collision is the real work. These branches add #[ignore]d measurement modules and main has since added its own in the same file regions. #1087's runtime.rs carries a single 456-vs-668-line hunk where its telemetry harness meets main's #1106 congruence probe, with near-duplicate helpers under different names. ~1,100 lines of test-only reconciliation per branch: it cannot break production, but it decides whether a published measurement reproduces. Also records that the five phase3 branches fork from a common merge-base with main but none is an ancestor of another — five independent experiments, no single resolution that templates them. #1109 is resolved and now merges clean; its story is kept as the template for job (a). Its alarming ~890-line runtime.rs conflict was both sides appending an independent module at EOF. The genuine work was porting its harness off the deleted env_extraction_policy onto Optimizer + Budget::Explicit, which keeps the caps as parameters the A/B needs. #1054's entry is rewritten from "recommend closing" to "fixed, rerun mutants then merge" — it was red and is now building, with the four deleted tests verified as targeting code that no longer exists. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NV5ZT2mxTnBC66unvBE7FF

#1109 reconciled, #1054 fixed and current, #994 brought to behind=0. Six of fifteen now merge cleanly and the one red CI is green. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NV5ZT2mxTnBC66unvBE7FF

Records the structural finding this day's work produced, which matters more than any individual merge: reconciling a branch against the saturation seam has a half-life measured in hours. Evidence. The board went 5/15 clean to 7/15 as #1054, #1109 and #1087 were fixed, then back to 5/14 the moment #1087 merged — its landing re-conflicted #1109 and #1114, both clean an hour earlier. #1109 needed reconciling twice in one day for the same end-of-file append collision. #1114 was clean at 10:10 and needs an API port by 11:00 (Budget gained a field, SaturationStop::ClassCap changed arity, reused helpers went private). And the #1087/#1101 collision this document predicted materialized exactly as described. One mechanism behind all of it: egraph/graph.rs, egraph/saturate.rs and runtime.rs are one hot seam with ten branches queued on it, and each landing invalidates the rest. The fix is not more reconciliation — it is to rebase the queue as a batch after each seam landing, or freeze the seam until it drains. Corollary: reconcile in landing order and land promptly, because a reconciled-but-unlanded branch is a wasting asset. #1087 is the counter-example in the good direction, reconciled and merged the same hour. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NV5ZT2mxTnBC66unvBE7FF

Pulls the calls scattered through the document into one table: merge #1054 (after a mutants rerun) and #1072; rework #1109 rather than merging it as written, since its change is unbounded on a clock-free main — 353 seconds for a 279-node kernel — and should be paired with Budget::Applications and re-measured, with its stop-re-arming half landable separately; review #1113 and #1114 normally; decide #994 either way; hold the seven gated on the class-cap question. Records the superseded set for the record: #1050 and #1044, both closed and neither salvageable. Notes that #1054 looked like a third case and was not — it was twice-invalidated but its tests were re-targetable, and it is now green. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NV5ZT2mxTnBC66unvBE7FF

claude added 3 commits August 31, 2026 11:35

docs: journal the 2026-08-31 test-quality audit (x86_64.rs mutation g…
3d1b581
…aps) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BR5xrsVNGXHBceptcxBvui

jppittman deployed to ci August 31, 2026 11:47 — with GitHub Actions Active

jppittman marked this pull request as ready for review September 1, 2026 02:14

chatgpt-codex-connectorBot reviewed Sep 1, 2026
View reviewed changes

Comment threaddocs/bugs/2026-08-31-test-quality-audit-followup.md Outdated

chatgpt-codex-connectorBot reviewed Sep 1, 2026
View reviewed changes

Comment threaddocs/bugs/2026-08-31-test-quality-audit-followup.md Outdated

jppittman changed the title ~~test(pixelflow-codegen): close x86_64.rs mutation gaps, remove dead code~~test(pixelflow-codegen): close x86_64.rs mutation gapsSep 1, 2026

merge main into claude/zen-babbage-wjmnit for#1054(unrelated nnue c…
9b2f8d2
…ommit)

Merge branch 'main' into claude/zen-babbage-wjmnit
136cc63

Merge branch 'main' into claude/zen-babbage-wjmnit
d988c5c

jppittman mentioned this pull request Sep 1, 2026
docs(results): open-PR triage — rebase sweep, blockers, and closure recommendations #1086
Open

This was referenced Sep 1, 2026
docs(results): open-PR sweep follow-up — merges executed, board corrected #1089
Merged
docs(plans): dead code with decent ideas — keep / reuse-for-round-2 / segregate / delete inventory #1090
Merged

Merge branch 'main' into claude/zen-babbage-wjmnit
a42a16d

jppittman mentioned this pull request Sep 4, 2026
test(pixelflow-codegen): close analyze_select_guards mutation gaps #1154
Open
6 tasks

Navigation Menu

test(pixelflow-codegen): close x86_64.rs mutation gaps - #1054

test(pixelflow-codegen): close x86_64.rs mutation gaps#1054
jppittman wants to merge 11 commits into
mainfrom
claude/zen-babbage-wjmnit

jppittman commented Aug 31, 2026•
edited
Loading

Uh oh!

github-actionsBot commented Aug 31, 2026

Uh oh!

chatgpt-codex-connectorBot left a comment

Uh oh!

Uh oh!

chatgpt-codex-connectorBot left a comment

Uh oh!

Uh oh!

chatgpt-codex-connectorBot commented Sep 1, 2026

Uh oh!

chatgpt-codex-connectorBot commented Sep 2, 2026

Uh oh!

jppittman commented Sep 2, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

jppittman commented Aug 31, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Test plan

Uh oh!

github-actionsBot commented Aug 31, 2026

Uh oh!

chatgpt-codex-connectorBot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

chatgpt-codex-connectorBot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

chatgpt-codex-connectorBot commented Sep 1, 2026

Uh oh!

chatgpt-codex-connectorBot commented Sep 2, 2026

Uh oh!

jppittman commented Sep 2, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

jppittman commented Aug 31, 2026•
edited
Loading