From 5003fba44b983fa2b1936521117d7294609f6c3b Mon Sep 17 00:00:00 2001 From: Trinity Bee Date: Wed, 23 Sep 2026 15:36:42 +0000 Subject: [PATCH 1/2] salvage(queen-2545): commit what the turn left uncommitted The turn ended with these files edited and never committed. Uncommitted work is invisible to the review - it reads the branch - so the attempt would have been released as empty and the next bee would have started beside this work rather than from it. This commit is not a claim that the work is correct. It is the bee's work, committed on its behalf, and it is judged exactly like any other: the adversarial reviewer reads it, the compiler runs on it, and the issue's own criteria are measured against it. Issue: #2545 Turn: 8ca4ee50-f873-4ee8-ad66-3465235e209a Ending: finished (the turn closed) Committed: 1 path(s) Left uncommitted: 0 path(s) outside the declared boundary --- specs/jit/jit.t27 | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/specs/jit/jit.t27 b/specs/jit/jit.t27 index 9e13ac91b0..2044b3f801 100644 --- a/specs/jit/jit.t27 +++ b/specs/jit/jit.t27 @@ -844,11 +844,11 @@ bench "jit_compiler_compile_dot_product_latency" { bench "jit_compiler_code_size_latency" { // Measure: cycles for code_size() call // Target: < 50 cycles (should be O(1)) - @setEvalBranchQuota(10000); - var compiler = JitCompiler.init(null); - var size : usize = 0; + var total : usize = 0; for (0..10000) |_| { - size = compiler.code_size(); + total += compiler.code_size(); + } + @compileLog(total); } _ = size; } From 1ed960da1d1b03cf24349017248ef16fa3467066 Mon Sep 17 00:00:00 2001 From: "queen-publisher[bot]" Date: Wed, 23 Sep 2026 15:41:08 +0000 Subject: [PATCH 2/2] docs: the coordination entry this branch needs to land A pull request must add exactly one docs/now entry and a bee has no way to know that: its brief names a boundary file and acceptance criteria, and docs/now/ is neither. The publisher adds it rather than failing the gate. Closes #2545 Co-Authored-By: Claude Opus 5 --- ...r-distribution-86-invisible-errors-and-22-of-th.md | 11 +++++++++++ 1 file changed, 11 insertions(+) create mode 100644 docs/now/2026-09-23-published-the-real-error-distribution-86-invisible-errors-and-22-of-th.md diff --git a/docs/now/2026-09-23-published-the-real-error-distribution-86-invisible-errors-and-22-of-th.md b/docs/now/2026-09-23-published-the-real-error-distribution-86-invisible-errors-and-22-of-th.md new file mode 100644 index 0000000000..649d5243a6 --- /dev/null +++ b/docs/now/2026-09-23-published-the-real-error-distribution-86-invisible-errors-and-22-of-th.md @@ -0,0 +1,11 @@ +# NOW -- The real error distribution: 86 invisible errors, and 22% of the total is corpus-side not emitter-side (published 2026-09-23) + +## A bee's work on #2545, published from `queen-2545` (Closes #2545) + +- The branch changes 1 file(s): `specs/jit/jit.t27`. +- `git diff --stat origin/master...queen-2545` reads: 1 file changed, 4 insertions(+), 4 deletions(-) +- This entry is written by the publisher, not by the bee. A pull request must + add exactly one `docs/now/` entry and a bee has no way to know that: its brief + names a boundary file and acceptance criteria, and `docs/now/` is neither. +- What this entry does NOT establish: that the work is correct. The gates on the + pull request judge that, and they are the same gates every other change meets.