apps: registry gains default=exclude; app-picker gains --installed and --audit - #89
Draft
yulonglin wants to merge 4 commits into
Draft
apps: registry gains default=exclude; app-picker gains --installed and --audit#89yulonglin wants to merge 4 commits into
yulonglin wants to merge 4 commits into
Conversation
…d --audit The Brewfile listed 40 entries while this Mac carried 53 casks and 20 App Store apps, and nothing compared the two. Now: - config/apps.conf: `default=exclude` for apps considered and rejected (cursor, codex-app, codexbar, conductor, espanso, aqua-voice); 15 casks and 12 App Store apps that were installed around the registry get rows. - app-picker --installed preselects from machine state; --audit compares the Mac against the Brewfile and prints uninstall commands, unregistered casks and App Store apps (row template with iTunes id lookup), formulae declared nowhere, and third-party taps. App Store apps are detected from _MASReceipt files because `mas list` hangs here (mas 7.0.0, macOS 26). Never proposes `brew bundle cleanup`, which would remove every config.sh formula. - mas-get and auth-setup skip exclude rows. - tests/test_app_picker.zsh: 28 checks with a stubbed brew and a fake /Applications. - Brewfile regenerated with --installed: 65 entries. - artifacts/brewfile-audit: the triage report, published. Claude-Session: https://claude.ai/code/session_01NVCGydT8aREk4q6VvzeoBy
…ection Codex review of PR 89 found two real defects and one hardening: - app-picker --audit built its registry lookup with a quoted associative subscript, which zsh can keep literally, so no Brewfile line matched. Key goes through a variable now; the test asserts a selected+installed app draws no uninstall command and that a hand-added missing cask is reported. - mas-get walked the whole registry and acquired every non-excluded App Store row, ignoring what app-picker deselected. It now reads the Brewfile's mas lines (--file), falling back to default=true rows when there is no Brewfile. Tested with a stubbed mas. - An empty inventory on a fresh Mac no longer short-circuits the audit; only a missing brew does. Report headline reworded: four of the six excluded casks were never in the Brewfile, so "leave the Brewfile" overstated the diff. Claude-Session: https://claude.ai/code/session_01NVCGydT8aREk4q6VvzeoBy
…sanctioned taps
- apps.conf: cloudflare-warp default=exclude ("for now"); visual-studio-code
default=true; LibreOffice note records that no tool here shells out to it.
- config.sh: PACKAGES_TRIAL_MACOS for tools installed by hand under
evaluation (hunk, codersauce/tap/red; micro as a commented candidate).
install.sh does not read it; app-picker --audit counts it as declared.
- app-picker: a third-party tap whose installed formulae are all declared
is listed as an exception on record instead of a policy breach. Test
covers both kinds of tap.
- Brewfile regenerated with --installed: 64 entries.
- Report revised with the review decisions, the Codex/ChatGPT answer, the
LuLu and LibreOffice answers, and a table tracing every library formula
to the requested tool that needs it (79 formulae, 39 requested, none
orphaned).
Removed on this Mac during review, not in the diff: marp-cli, graphite
(+untap), age, sops; ChatGPT Classic.app to Trash.
Claude-Session: https://claude.ai/code/session_01NVCGydT8aREk4q6VvzeoBy
- app-picker --deps prints why every installed formula is here: requested (and where it is declared: apps.conf, config.sh, helpers.sh, nowhere) or pulled in, traced to the requested leaves through runtime_dependencies. Every Brewfile write and every --audit rewrites docs/brew-formulae.md, so the doc refreshes whenever package state is being looked at; no timer, because this repo deploys no user launchd jobs and a scheduled rewrite would leave uncommitted drift. CLAUDE.md links it in one clause; nothing loads it into agent context. - config.sh: `tldr` (deprecated C client) leaves PACKAGES_CORE, which also feeds apt; tlrc (official Rust client) joins PACKAGES_MACOS and PACKAGES_LINUX_BREW. The audit no longer flags tlrc. - test: stubbed `brew info --json=v2` fixture; 48 checks. - report: LuLu, provenance and tlrc answers. Claude-Session: https://claude.ai/code/session_01NVCGydT8aREk4q6VvzeoBy
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
What
The Brewfile listed 40 entries while this Mac carried 53 casks and 20 App Store apps, and nothing compared the two.
config/apps.conf:defaultgains a third value,exclude, for apps considered and rejected. Seven rows carry it (cursor, codex-app, codexbar, conductor, espanso, aqua-voice, cloudflare-warp). Fifteen casks and twelve App Store apps that were installed around the registry get rows; defaults follow whether they run, sit in login items, or were opened this fortnight, then Yulong's review.custom_bins/app-picker:--installedpreselects from machine state (deselecting a row is the UI for dropping an app);--auditcompares the Mac against the Brewfile and prints uninstall commands, unregistered casks, unregistered App Store apps with a row template, formulae declared nowhere, and third-party taps split into breaches and exceptionsconfig.shdeclares. Runs after every write too. Never proposesbrew bundle cleanup, which would remove everyconfig.shformula.custom_bins/mas-getacquires only the Brewfile's App Store lines (--file), falling back todefault=truerows without one. Before, it walked the whole registry and would have reacquired deselected apps.auth-setupskipsexcluderows.config.sh:PACKAGES_TRIAL_MACOS, tools installed by hand under evaluation that the audit counts as declared (hunk, red; micro as a candidate). install.sh does not read it.config/Brewfileregenerated with--installed: 64 entries.tests/test_app_picker.zsh: 41 checks with a stubbedbrewandmasand a fake/Applications.artifacts/brewfile-audit/).Done on this Mac during review (not part of the diff)
brew uninstall marp-cli graphite age sops,brew untap withgraphite/tap,brew autoremove(removed nothing), andChatGPT Classic.appmoved to the Trash.install_age/install_sopsin helpers.sh have no callers; left in place.Why App Store detection uses receipts
mas listhangs on this machine (mas 7.0.0, macOS 26; a 25 s timeout printed nothing), so the audit scansContents/_MASReceipt/receiptunder/Applicationsand matches registry rows by name: exact stem, else first word (Things 3↔Things.app).Review points
declared_elsewhereis agrep -wof the token againstconfig.shandscripts/shared/helpers.sh; it keeps thecodexandantigravity-clicasks out of the unregistered list and would also suppress a formula whose name happens to appear as a word in either file.https://claude.ai/code/session_01NVCGydT8aREk4q6VvzeoBy