stack/agents/filbert/work/queue-154/r3/body.md at dev-console

0 Fork 1

Files

T

jason.woltje b9ea9304fd lead: commit filbert's untracked evidence (273 files: queue-101 queue-102 queue-105 queue-107 queue-108 queue-109 queue-110 queue-112 queue-113 queue-114 queue-115 queue-116 queue-119 queue-120 queue-122 queue-123 queue-125 queue-126 queue-128 queue-129 queue-130 queue-131 queue-132 queue-133 queue-134 queue-135 queue-136 queue-137 queue-139 queue-140 queue-141 queue-142 queue-145 queue-146 queue-147 queue-148 queue-149 queue-150 queue-151 queue-152 queue-153 queue-154 queue-155 queue-156 queue-162 queue-163 queue-165 queue-166 queue-167 queue-169 queue-170 queue-175 queue-177 queue-181 queue-182 queue-184 queue-185 queue-186 queue-187 queue-188 queue-49 queue-64 queue-66 queue-70 queue-71 queue-80 queue-84 queue-85 queue-87 queue-88)

Unclaimed files left in the canonical checkout. Inventory: canonical-cleanup-inventory.md (commit-as-evidence groups). detect-secrets scan clean on 2026-11-02: hits are sha256 pins, git object ids, test sentinels already in tracked tests and unit-file lines.

2026-11-02 15:05:14 -06:00

3.9 KiB

Raw Permalink Blame History

Filbert review, row 154 round 3: approve

Candidate manifest cebca5145c899a362e938d36b26168070ad8cc01bfdc0c9f44a68eee07a005ca (195 paths), r3/candidate.patch (d709b2dd...) applied with --index to 850484a7d. The manifest matched before and after every run. Compared with r2, five paths changed: editor/BUILD.json, editor/model.mjs, src/public/editor.js, tests/docs-browser.test.mjs and tests/editor-browser.test.mjs.

My r2 change is fixed, and so are my non-blocking notes. No new findings.

Change from r2, fixed: `BUILD.json` records the shipped bundle

editor/BUILD.json:6-8 records src/public/editor.js as 9c881b13740fae2408717ad75fa1d85bfcfe3fa8b3af0f1b3f2e706c90e80fee, 758420 B. That is the shipped file.
I rebuilt with the pinned toolchain (lock eb60a494..., esbuild 0.28.2 binary e1698a3d..., toolchain check rc=0 under MOSAIC_REQUIRE_EDITOR_BUILD=1), both to a scratch --out and in place. In both, editor.js, editor.css (8654ac57..., 4462 B) and BUILD.json are byte for byte the same as the shipped files.
editor-provenance.test.mjs passes 6/6 under MOSAIC_REQUIRE_EDITOR_BUILD=1.

r2 notes, checked

Title backslashes.
- model.mjs:172: titleBackslash doubles each backslash before the vendored link writer escapes ", so x\") is written as x\\\") and end\ as end\\.
- :193: the image title escapes both characters in one pass (/[\\"]/g).
- editor-browser.test.mjs:481-488 adds four tests. They cover a quote after a backslash then a tag, and a trailing backslash, each for a link title and an image title. Each asserts the exact written string, then re-reads the saved text with a read-only mount (:454-465).
g5. :492 pastes <pre data-params="ab">` and requires a bare fence on Save. My g5 is now killed by it.
docs-browser.test.mjs:67. The explicit document.querySelector("#doc-commit") check is back, as at :65.
Tilde wording. BUILD.md :419 and :464 now say a raw source block (pre.md-raw) inside the visual editor.

Non-blocking note

The re-read asserts haven't been shown to fail on their own. BUILD.md :490 says this itself: on every backslash mutant (Researcher's and my g1, g8 and g9) the exact-string assert fails first. I read :454-465, and the re-read checks the right things: the parsed title equals the pasted value, and no <img appears in the text. The exact-string assert already pins each written form, so the tests kill a wrong escape either way. I don't need that extra run.

Ran by me

Slot from sage-0858. The busy check was CLEAR before each step and after the last one. TMPDIR was in my own scratch, DOCKER_HOST was a dead socket, and MOSAIC_REQUIRE_EDITOR_BUILD=1 was set.

Rebuild, 19:09:28Z to 19:09:29Z. Results as above. The shipped files were restored afterwards and the manifest re-matched.
Singles, 19:09:35Z to 19:12:48Z.
- Every packages/webui/tests/*.test.mjs file was run singly: 32 files, 206/206 by awk sum, 0 skipped.
- That includes editor-browser 16/16, editor-provenance 6/6, editor 5/5 and docs-browser 1/1.
Mutants, 19:12:54Z to 19:15:58Z.
- Both controls (editor-browser.test.mjs, editor.test.mjs) passed first.
- Each model.mjs mutant was rebuilt into the bundle before its run, and the files were restored after each one.
- 9 run, 9 killed:
  - g1 (link title not escaped at all): tests 7, 11, 13
  - g2: test 8
  - g3: test 9
  - g4: test 10
  - g5: test 15
  - g6: test 16
  - g7 (docShown never set, bus.js): tests 4, 5, 6
  - g8 (link backslash not doubled): tests 11, 13
  - g9 (image title escapes " only): tests 12, 14

Not mine

Researcher's r3 runs (BUILD.md "Round 3 results"): two byte-identical builds; singles 206/206; suite 206/206; 7 of 7 mutants killed; provenance 6/6 after the mutants; 195 of 195 manifest checks. I read the logs list and the mutant table, and did not rerun their script. My counts agree with theirs.

3.9 KiB Raw Permalink Blame History