mutation: the last four HOLEs, x/text 0.43.0, golangci-lint v2.15.0 by donislawdev · Pull Request #197 · donislawdev/TestingFilesGenerator

donislawdev · 2026-10-19T16:49:00Z

Three things in one pull request, one commit each.

The last four HOLEs of the Sunday mutation run (2026-10-18: 1782 entries, 1725 ok, 4 HOLE, 0 BROKEN, 0 SKIP). All four were about the width of a menu.

A field in the form's grid is as wide as the columns it takes, not as wide as it asks. So at the width the window opens at, taking the floor out of parts.menuWidth changed nothing a guard could see. The floor only shows in the narrowest window, where it decides how many columns a menu takes: without it, the damage menu on the single batch screen falls from 265.91 px to 124.96, beside boxes of 253.91.
inEveryMenuLayout lays every screen out with every section open, at the opening width and at the narrowest one. It asserts both states instead of assuming them. The guard about the toolkit's words also asks whether the width a menu asks for covers the toolkit's minimum. Laid out, the menu without its floor was still 0.73 px wider than the words.
Two entries are retired: measuring the placeholder in menuWidth. Since NewChooser gives every menu the window's placeholder from the start, that change widens every format menu at once (154.73 to 180.23 px), which is still one width everywhere. Their two guards are listed as not proven by mutation. The cause is held at the source by TestAMenuIsOneWidthBeforeAndAfterItIsDrawn.

golang.org/x/text 0.43.0, in place of #196, which was red only on the notices file. Only unicode/bidi and secure/bidirule changed between the two module copies, and neither binary links them, so no generated byte can move.

golangci-lint v2.15.0, from the weekly tool versions job. v2.15.0 refuses exhaustive.ignore-enum-types in config verify, so the setting is now ignore-types. Checked: without the setting the linter reports seven switches on fyne.KeyName, with it none. A full run reports no issues.

The merge needs the browser, because .github/workflows/ci.yml changes.

🤖 Generated with Claude Code

Summary by CodeRabbit

Maintenance
- Updated the project’s code-quality tooling and refreshed a supporting text-processing component.
Tests
- Expanded menu layout checks across screens, window widths, and open sections. Checks now verify that menus remain wide enough for their contents and aligned with nearby input fields.
- Added coverage for consistent menu widths across screens and table-row alignment.

…t window The last four HOLEs of the Sunday mutation runs were all about the width of a menu, and none of them was a guard looking in the wrong place for the reason the earlier analysis gave. A field in the form's grid is as wide as the columns it takes, not as wide as it asks. At the width the window opens at, every menu is a column wide whatever it asks for, so taking the floor out of parts.menuWidth changed nothing the guards could see. The floor only shows in the narrowest window the window allows, where it decides how many columns a menu takes: without it, the damage menu on the single batch screen falls from two columns (265.91 px) to one (124.96), beside boxes of 253.91. inEveryMenuLayout lays every screen out with every section open and at both widths, and asserts both states rather than assuming them: a section that stays shut when its head is pressed, or content that did not take the width it was resized to, stops the guard. The guard about the toolkit's own words also asks whether the width a menu asks for covers the toolkit's minimum - laid out, the menu without its floor was still given one column, 0.73 px wider than the words, so the layout alone could never prove it. The two entries that measured the placeholder in menuWidth are retired. Since NewChooser gives every menu the window's placeholder from the start, measuring it widens every format menu at once (154.73 to 180.23 px), which is one width everywhere. Their guards are listed as not proven by mutation, and the cause they were written for is held at the source by TestAMenuIsOneWidthBeforeAndAfterItIsDrawn. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

The Dependabot pull request (#196) was red on one guard: the notices file writes the version by hand and still said 0.42.0. Between the two copies in the module cache only unicode/bidi and secure/bidirule changed. Neither binary links either of them - go list -deps of cmd/tfg and cmd/tfg-gui reaches cases, language, transform and unicode/norm of this module, all four unchanged - so no generated byte and no verdict about two names colliding can move. The mutation entry that names the version in the notices moves with it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

Reported by the weekly tool versions job. v2.15.0 refuses exhaustive.ignore-enum-types in config verify, which CI runs before the linter, so the setting is ignore-types now, a list of the same patterns. Checked rather than assumed: without it the linter reports seven switches on fyne.KeyName in internal/gui/parts, with it none. A full run on the tree reports no issues. The two mutation entries that name the pinned command move with it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

coderabbitai · 2026-10-19T17:05:38Z

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Repository UI (base), Organization UI (inherited)
Review profile: ASSERTIVE
Plan: Advanced
Run ID: 803ad706-f43f-44e2-86f7-2f2a9c7d4d74

📥 Commits

Reviewing files that changed from the base of the PR and between fd6a3eb and f05048a.

⛔ Files ignored due to path filters: 1

go.sum is excluded by !**/*.sum, !**/go.sum

📒 Files selected for processing: 8

.github/workflows/ci.yml
.golangci.yml
THIRD-PARTY-NOTICES.md
go.mod
internal/guard/menushape_test.go
internal/guard/mutation/entries.py
internal/guard/mutationcoverage_test.go
internal/gui/parts/ring.go

Included review availability: This review used your included allowance. Your plan provides up to 1 included review per hour; 0 remain after this review.

📜 Recent review details

⏰ Context from checks skipped due to timeout: 3

GitHub Check: race detector (part 1 of 4)
GitHub Check: race detector (part 0 of 4)
GitHub Check: race detector (part 2 of 4)

🧰 Additional context used

📚 Code guidelines: 1

SECURITY.md — unknown

📓 Path-based instructions: 15

For every added or upgraded dependency: confirm the package really exists and the name is spelled correctly (typosquatting), it is actively maintained, the license is compatible with this project's license, and it is actually needed (not re...