[tnalbant]

What this changes, and why

Six user-visible counts were written with a plural noun. Each now reads the singular at exactly one:

• the panel's whitespace row: Whitespace only · 1 character (Clip.swift)
• the Insights range label: 1 word · 16 October, and the coverage line: Dictated on 1 of 1 day shown. (InsightsPresentation.swift)
• three Diagnostics counters: 0 of 1 piece, 0 of 1 segment, 0 of 1 dictation (DiagnosticsPresentation.swift)
• the typing-stopped error: Typing stopped after 0 of 1 character. (TextInsertionError.swift)

The obvious inline fix (count == 1 ? "character" : "characters" at each site) would have repeated the same ternary in six files across three modules, and UttrflowClipboard and UttrflowCore cannot import UttrflowUX where the existing MainFormatting.count lives. So CountedNoun goes into UttrflowCore — of(_:singular:plural:) for the noun alone and count(_:singular:plural:) for the number with it — and MainFormatting.count now delegates to it, so the surfaces cannot drift apart again.

The noun follows the number the sentence actually prints: the Insights coverage line counts reachable days, so 1 of 2 days shown stays plural, and 1 of 1 day shown does not.

How you know it works

Unit cases added, one per surface: CountedNounTests (new file), a ClipTests case at one and two spaces, InsightsPresenter.caption and .coverage at a count of one and of two, and DiagnosticsPresenter.decodingRows / .segmentRows at a single piece and a single segment (both also at the plural stay unchanged).

The test suite did not run on my machine, and I have not ticked make verify below. swift test cannot build the test bundle here, on two failures that are present on a clean checkout of main and in files this change does not touch:

• Tests/UttrflowClipboardTests/SecretShapesMutationTests.swift and SecretShapesOracleTests.swift: the compiler is unable to type-check this expression in reasonable time
• Tests/UttrflowInputTests/SuggestionAcceptorTests.swift: the compiler crashes while emitting SIL for insertRefusesWhenCancelled

Both are Swift 6.4 / macOS 27 toolchain behaviour; CI runs macos-26. Moving those files aside does not help, because other tests in the same target use helpers they define.

What did run locally:

swift build --target UttrflowCore        → 0 errors
swift build --target UttrflowClipboard   → 0 errors
swift build --target UttrflowUX          → Build complete
swift build --target UttrflowCoreTests   → Build complete
swift build --target UttrflowUXTests     → Build complete
make lint                                → clean (LineLength/formatting fixed with make format first)
make comment-audit                       → 0 multi-line blocks, baseline unchanged
make verify-static                       → green (includes public-api-audit after the baseline update below)

So the new test files compile, but they have not executed anywhere yet: in .github/workflows/ci.yml a pull request runs the static job only (make verify = verify-static), and the full gate (make verify (full): build, unit tests, coverage, offline audit) runs on workflow_dispatch. A maintainer running it will be the first execution; if it fails there, it is mine to fix.

The change adds one public type and two public functions to UttrflowCore, so Scripts/public_api_baseline.json records them (enum CountedNoun: 1, func count: 2, func of: 7) — the second commit is that line, so the audit reviews the delta instead of failing on it.

---

• make verify passes locally — the static half does (make verify-static is green here, and the make verify job is green on this head); the build the other half needs cannot complete on this machine (lint, PII audit, build, tests, coverage floor, offline audit) — see above: build and tests cannot run on this machine, lint and the comment audit do pass
• New behaviour has a test, or there is a reason in the PR why it cannot
• No real email address, postal address or personal data in fixtures — no fixtures added
• Comments are one line, present tense, and describe what the code does now
• A latency claim says where its clock starts and stops — no latency claim here

Drafted with an AI coding assistant (OpenCode Go, deepseek-v4.1-flash) and reviewed against the issue's expected strings; the runs quoted above are from this branch.

Fixes #7913

[greptile-apps]

Your trial has ended. Reactivate Greptile to resume code reviews.

[coderabbitai]

Important

• Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

Run configuration

• Source mode: Auto
• Repository: uttrflow/uttrflow-swift
• Branch: main
• File: .coderabbit.yaml
• Commit: 8483081
• Layers: Repository YAML
• Review profile: ASSERTIVE
• Plan: Advanced
• Run ID: 8af3b612-6f02-4be2-8b57-c166da8484c7

• 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.

Share

• X
• Mastodon
• Reddit
• LinkedIn

[greptile-apps]

Your trial has ended. Reactivate Greptile to resume code reviews.

[greptile-apps]

Your trial has ended. Reactivate Greptile to resume code reviews.

[tnalbant]

Rebased onto main (8483081, release 26.1106.0). Two things changed in the rebase.

The formatted-text row is gone. PanelPresenter.measurements no longer renders a rich-text row on main: upstream removed that branch, so the surface this PR fixed there no longer exists. I dropped that half of the change and the test that went with it rather than keep a call into a row nothing shows. Six surfaces remain: the whitespace row, the Insights range label and its coverage line, the three Diagnostics counters, and the typing-stopped sentence.

The baseline was re-recorded by the repo's own tool. Scripts/public_api_audit.py --update --after-merge writes the three declarations CountedNoun adds (the enum, and the count and of overloads), so Scripts/public_api_baseline.json shows 9073 -> 9076 in the diff instead of failing the audit.

Local evidence on this Mac:

• swift build --target UttrflowUX builds.
• swift build --target UttrflowCoreTests builds, and running UttrflowCoreTests.xctest executes 1219 tests in 175 suites with the new CountedNoun suite passing. The 18 issues that run reports are all TimestampsTests (Yesterday 4:14 PM, Wed, 1 Jul at 12:46 PM), in a file this change does not touch.
• swift build --target UttrflowUXTests and swift build --target UttrflowClipboardTests still cannot build here: the first stops on Sources/UttrflowAI/AppleFoundationCleanupModel.swift:19 (GenerationOptions(sampling:) is deprecated against the macOS 27 SDK and the target compiles with warnings as errors), the second on a type-check budget failure inside a @Test macro. Both fail on a clean main on this machine too.

So the Insights and Diagnostics unit cases are unverified locally; CI is the gate for those.