You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
b39fd97 でローカルのゲートを通しました（gate.py、全段 ）。test は 1 回、無関係な Mach loopback の a failed register gives back the receive right it allocated（プロセス内の受信権の増減を数えるで、並列に走る他ののと思われますが、原因は確かめていません）で落ち、make test の再実行とゲートの通し直しはどちらも通っています。
…the cap bounds
The cap on the hover's description was documented as keeping a tooltip from being the size of a file, which it did not: it counted Characters (one letter under a thousand combining marks passed), trimmed before cutting (so padding could push a 280-character description over, and a cut could leave a gap before the ellipsis) and re-read the whole text on every keystroke. It now counts Unicode scalars, reads only the start of the text, ends in the ellipsis only when more than whitespace follows the cut, and the comment, README and ARCHITECTURE say that it bounds the description alone.
…ts entry carries
help(of:) already ignored a broken package's description, but artwork(of:) relied on the catalog never handing one a thumbnail, and its test used a fixture that had none. It now gates on availability itself and the test gives the broken entry a thumbnail. A description-only listing is now pinned as a new state at the UseCase and as a new hover at the Presenter, as a picture-only one already was.
… and to pixels-not-points
EffectThumbnailDecodingTests decoded only synthetic 8-bit grey bitmaps and cut a picture inside its header. They now also decode Examples/Ripple.lens/thumbnail.png (256 x 256, RGB) and check that it is not one flat colour, check that a solid colour comes back as that colour in both formats, pin that a zero-size PNG is nil, and judge the 2048 x 2048 limit by pixels with a picture whose points are half its pixels. The truncation test says what it pins (a cut inside the header); a cut inside the pixel data decodes to a partly drawn image on this OS, which is recorded in the suite and in ARCHITECTURE rather than held to. The decoder comment gives the measured cost of an over-limit file.
…here the Router holds a decode
ARCHITECTURE quoted a 1.7 MB PNG as the case the 2048 x 2048 limit guards against, which the catalog's 1 MiB limit would refuse first; it now quotes a 15 000 x 15 000 PNG of 982 742 bytes with its measured open and draw costs. The Known limits paragraph said a thumbnail is small and that the decode repeats on every body pass; it now gives the counted behaviour (once per built cell) and the real worst case, and ImageIO is stated as a choice and not as an import-list rule. A picture cut off inside its pixel data decodes partly and is written down as such, and that an undecodable or over-limit picture is not logged is stated in README and ARCHITECTURE. The Router row of architecture-boundaries.md, the chain note in CLAUDE.md, AGENTS.md, the Entity comment and the Package.swift label now name the picker's thumbnail decode and its pixel limit, and Lane 1 gets the command that makes its corrupt-PNG file.
No actionable comments were generated in the recent review.
Walkthrough
The picker now uses artwork only for ready entries and limits trimmed descriptions to 280 Unicode scalars. Thumbnail decoder tests cover image rendering and rejection cases. Documentation describes the decoder’s pixel limit, fallback behavior, and related picker behavior.
Documentation describes the 2048 × 2048 pixel limit, decode failure fallback, partial image rendering, and decoder responsibilities. Decoder tests cover PNG and JPEG color preservation, the shipped Ripple thumbnail, truncated headers, zero dimensions, and pixel dimensions versus point dimensions.
Artwork selection now uses thumbnails only for ready entries. Descriptions are trimmed and limited to 280 Unicode scalars, with an ellipsis when additional non-whitespace text follows. Tests cover description display updates and catalog state changes without additional preparation. The version changes to 0.24.0.
GeneralD/mirage#65: Adds the effect picker and the contracts used by the picker changes in this PR.
GeneralD/mirage#59: Introduces .lens manifest descriptions, which this PR trims and displays.
GeneralD/mirage#56: Implements catalog state publication, which the description-update test exercises.
Merge Risk: Low · previous review abbca
Search may lag when a package has an unusually long description ending in whitespace. This is a bounded merge risk, not a reason to block the release.
Pre-merge checks: 4 · 1
Failed checks: 1 warning
Check name
Status
Explanation
Resolution
Docstring Coverage
Docstring coverage is 66.67% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff with cyclomatic complexity of at least 2. Documenta…
Write docstrings for the functions missing them to satisfy the coverage threshold.
Passed checks: 4 passed
Check name
Status
Explanation
Description Check
Check skipped - CodeRabbit’s high-level summary is enabled.
Title check
The title accurately identifies the picker review follow-up and the version bump to 0.24.0, which are the main changes in the pull request.
Linked Issues check
Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check
Check skipped because no linked issues were found for this pull request.
Full details: Docstring Coverage
Explanation
Docstring coverage is 66.67% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff with cyclomatic complexity of at least 2. Documentation-only edits and deletions within surviving functions are included; unchanged diff context and fully deleted functions are excluded. Analyzed 6 functions across 8 files. (8 skipped: 8 unsupported.)
Resolution
Write docstrings for the functions missing them to satisfy the coverage threshold.
Finishing Touches
Generate docstrings
Commit to this branch
Create a new PR
Generate unit tests (beta)
Commit to this branch
Create a new PR
Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.
Included review availability: This review used your included allowance. Your plan provides up to 1 included review per hour; 0 remain after this review.
The reason will be displayed to describe this comment to others. Learn more.
🟡 Changes recommended
Two active issues remain: an integer-overflow crash is possible when validating untrusted thumbnail dimensions, and description matching can still scan unbounded whitespace-padded suffixes.
This follow-up addresses post-merge review findings from #69, improves effect-picker descriptions and thumbnail handling, expands regression coverage, updates documentation, and bumps the version to 0.24.0.
Changes:
Truncates descriptions by Unicode scalar count while preserving full names and authors.
Restricts thumbnail artwork to ready entries and expands decoder/catalog test coverage.
Synchronizes architecture, testing, and verification documentation.
Updates version.txt from 0.23.0 to 0.24.0.
Open findings remain around overflow-safe pixel-limit handling and unbounded whitespace scanning.
pixelsWide * pixelsHigh traps in Swift when a header claims sides near UInt32.max, so a crafted thumbnail could crash the picker instead of drawing the tile. The limit is now width <= maxPixels / height, with zero sides refused in the same guard.
The ellipsis check scanned the rest of the description until a non-whitespace scalar, so a description padded with whitespace after the cut was read through on every keystroke. It now looks at no more than 280 scalars past the cut, which the comment, ARCHITECTURE and a test case state, including that a longer gap hides what follows it without an ellipsis.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Follow-up of #69
NSBitmapImageRepのfeat:の feat(picker): ピッカーのセルにマニフェストのサムネイルを描き、ホバーに説明を加える #69 が上げ損ねた minor を0.24.0として上げます（chore(release)の 1 コミット）。Character数で数え（artwork(of:)が可用性を見ていなかった（nit、7d8ea69）:help(of:)は broken のavailability == .readyを見るようにし、EffectCatalogThumbnailTests）と Presenter（ホバーが変わる）の両方で固定しました。Examples/Ripple.lens/thumbnail.png（256 × 256 RGB）を実際にデコードして一色でないことをnilであること、解像度（pt）が半分の画像でも画素数でnil）を足しました。「途中で切れた」testing.mdの読み取り例外に、2 つ目の読み手としてEffectThumbnailDecodingTestsを明記しました。architecture-boundaries.mdの Router 行、.claude/CLAUDE.mdのチェーン注記、AGENTS.md（2048 × 2048 のEffectThumbnailのコメント、Package.swiftの Router 行を、printfコマンドを足しました。EffectArtworkResolutionTests.nothingIsRememberedはキャッシュの有無を約束するように読めたので、「リゾルバは何も覚えない」という意味の名前に直しました（ee92e9e）。pixelsWide * pixelsHighは、ヘッダがUInt32.max近い辺を宣言するとトラップするので、積を使わずwidth <= maxPixels / heightで判定し、0 の辺もversion.txtを 0.23.0 から 0.24.0 に上げました（chore(release)）。NSHostingViewにデコーダの代わりのカウンタを置き、絵のあるセル 9 つ + タイルのセル 2 つで数えたところ、最初の描画が 9 回、LazyVGridから出入りしたときや画像が変わったとき）だけです。コストが見えていないものに、キャッシュか第二の経路を足さないことにしました。ARCHITECTURE の Known limits に、数えたNSBitmapImageRep(data:)が先にデコードしているからarchitecture-boundaries.mdなどに明記）を採っています。NSBitmapImageRepがnilを返すので、nilにする（should-fix）: 切れた位置から先が黒や灰で描かれるIENDなら確実ですが、JPEG は終端マーカーのあとに余りがあっても.claude/CLAUDE.mdの「SwiftUI の view body は unit test しない」というルールを変えるmaxPixelsの判定とそのb39fd97でローカルのゲートを通しました（gate.py、全段testは 1 回、無関係なMach loopbackのa failed register gives back the receive right it allocated（プロセス内の受信権の増減を数えるmake testの再実行とゲートの通し直しはどちらも通っています。format: PASS（lint: PASS/check: PASS/build: PASS/build-dev-harness: PASStest: PASS—Test run with 1052 tests in 139 suites passed ... with 12 known issues（12 件は main に元からある既知の issue で、このtest-gpu: PASS—Test run with 27 tests in 6 suites passedEffectPickerPresenterTests（余白 3 通り、EffectCatalogThumbnailTests（EffectThumbnailDecodingTests（NSBitmapImageRepの