config: drop settings that restate a tool default #264

Merged

natsukium merged 4 commits from chore/trim-default-config into main 2026-09-12 17:55:07 +09:00

Conversation 0 Commits 4 Files changed 108 +7 -211

natsukium commented 2026-09-12 17:44:26 +09:00

Owner

Copy link

A sweep over the repository's tool configuration for settings that say
nothing: values identical to the tool's own default, a duplicated MSRV,
advisory ignores whose advisories are gone, and lint allows the lint
policy already grants. Each removal was verified against the tool rather
than against documentation.

What goes

Dead advisory ignores. wayland-scanner no longer pins quick-xml
^0.39, so RUSTSEC-2026-0194 and -0195 never reach the graph; cargo-deny
was emitting an advisory-not-detected warning for each.

The duplicated MSRV. Clippy reads [workspace.package] rust-version
by itself, and prefers clippy.toml's msrv silently when the two
differ — so the duplicate was a second place to edit that could outvote
the manifest.

Settings equal to the tool's default: rustfmt's max_width and
tab_spaces; cargo-deny's db-urls, confidence-threshold,
[licenses] version and [bans] multiple-versions; buf's STANDARD
lint category. Removing all of them leaves cargo deny check output
byte-identical and treefmt a no-op. rustfmt's edition is redundant
too (treefmt passes --edition 2024 on the command line, which wins)
but is kept so a bare rustfmt <file> does not fall back to 2015 rules.
newline_style = "Unix" stays: it is the only real deviation there, and
the Windows checkout depends on it.

104 module-level #![allow(clippy::unwrap_used / expect_used)].
clippy.toml already exempts every item in a bare #[cfg(test)]
module — helper functions included, not just #[test] ones.

What deliberately stays

Clippy keys on the bare cfg(test), so the eight allows under
#[cfg(all(test, unix))] and #[cfg(all(test, feature = "schema"))]
modules are load-bearing, as are those in benches/** (never a test
context) and in tests/*.rs, where the helpers around the #[test]
functions are not exempt. Where an allow also named float_cmp or
cast_possible_wrap, only the in-test lints were dropped.

Verification

just check in full — fmt-check, lint, check-portable,
prose-check, skill-check, test (3489 passed), deny,
proto-compat-test, proto-compat, release-check-test — plus
nix flake check, which rebuilds the MSRV derivation.

A sweep over the repository's tool configuration for settings that say nothing: values identical to the tool's own default, a duplicated MSRV, advisory ignores whose advisories are gone, and lint allows the lint policy already grants. Each removal was verified against the tool rather than against documentation. ## What goes **Dead advisory ignores.** wayland-scanner no longer pins quick-xml ^0.39, so RUSTSEC-2026-0194 and -0195 never reach the graph; cargo-deny was emitting an `advisory-not-detected` warning for each. **The duplicated MSRV.** Clippy reads `[workspace.package] rust-version` by itself, and prefers `clippy.toml`'s `msrv` silently when the two differ — so the duplicate was a second place to edit that could outvote the manifest. **Settings equal to the tool's default:** rustfmt's `max_width` and `tab_spaces`; cargo-deny's `db-urls`, `confidence-threshold`, `[licenses] version` and `[bans] multiple-versions`; buf's `STANDARD` lint category. Removing all of them leaves `cargo deny check` output byte-identical and `treefmt` a no-op. rustfmt's `edition` is redundant too (treefmt passes `--edition 2024` on the command line, which wins) but is kept so a bare `rustfmt <file>` does not fall back to 2015 rules. `newline_style = "Unix"` stays: it is the only real deviation there, and the Windows checkout depends on it. **104 module-level `#![allow(clippy::unwrap_used / expect_used)]`.** `clippy.toml` already exempts every item in a bare `#[cfg(test)]` module — helper functions included, not just `#[test]` ones. ## What deliberately stays Clippy keys on the *bare* `cfg(test)`, so the eight allows under `#[cfg(all(test, unix))]` and `#[cfg(all(test, feature = "schema"))]` modules are load-bearing, as are those in `benches/**` (never a test context) and in `tests/*.rs`, where the helpers around the `#[test]` functions are not exempt. Where an allow also named `float_cmp` or `cast_possible_wrap`, only the in-test lints were dropped. ## Verification `just check` in full — `fmt-check`, `lint`, `check-portable`, `prose-check`, `skill-check`, `test` (3489 passed), `deny`, `proto-compat-test`, `proto-compat`, `release-check-test` — plus `nix flake check`, which rebuilds the MSRV derivation.

natsukium added 4 commits 2026-09-12 17:44:26 +09:00

deny: drop the advisory ignores no longer matching any crate ... 361417fd9e

wayland-scanner has moved off the quick-xml ^0.39 pin, so neither
RUSTSEC-2026-0194 nor -0195 reaches the graph any more. cargo-deny
reports an `advisory-not-detected` warning for each stale entry, which
under `-D warnings` would be a gate failure waiting to happen and in the
meantime trains readers to ignore the tool's output.

Assisted-by: Claude Code:Opus 5

lints: let the manifest's rust-version be the only MSRV ... ce78c65f3a

Clippy reads `[workspace.package] rust-version` on its own, so the
duplicate `msrv` key only created a second place to edit. When the two
drift clippy silently prefers clippy.toml, so a bumped manifest would
keep being linted against the old floor.

Assisted-by: Claude Code:Opus 5

config: keep only the settings that deviate from a tool's default ... 7fb45cb59b

rustfmt's max_width/tab_spaces, cargo-deny's advisory db URL, license
confidence threshold, `[licenses] version` and `[bans]
multiple-versions`, and buf's STANDARD lint category all restated the
tool's own default. Spelling a default out gives a reader no way to tell
a deliberate policy from an inherited one, and it silently pins the value
if the tool ever revises its default.

rustfmt's `edition` is redundant too — treefmt passes `--edition 2024` on
the command line, which wins over the file — but it is kept so a bare
`rustfmt <file>` does not fall back to 2015 rules.

Assisted-by: Claude Code:Opus 5

lints: stop re-granting in-test allows clippy.toml already grants ... b33f24b08d

`allow-unwrap-in-tests` / `allow-expect-in-tests` / `allow-print-in-tests`
exempt every item in a bare `#[cfg(test)]` module, helper functions
included, so the module-level `#![allow]` in each of them said nothing.
Dropping them leaves an `#![allow]` inside a test module meaning what it
should: this module needs something the test policy does not already
cover.

The allows under `#[cfg(all(test, …))]` modules stay: clippy keys on the
bare `cfg(test)`, so a compound gate is not a test context to it. So do
the ones in `benches/**` (never test context) and in `tests/*.rs`, where
only `#[test]` functions are exempt and the helpers around them are not.

Assisted-by: Claude Code:Opus 5

natsukium merged commit 0bebe11dbe into main 2026-09-12 17:55:07 +09:00

natsukium deleted branch chore/trim-default-config 2026-09-12 17:55:08 +09:00

natsukium referenced this pull request from a commit 2026-09-12 17:55:08 +09:00

config: drop settings that restate a tool default (#264)

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels   

priority/P0

Must be resolved before the v0.1.0 tag

  

priority/P1

Must be settled before the first public contract freezes

  

priority/P2

Explicitly deferred until after v0.1.0

  

release/v0.1.0

Work tracked for the first tagged release

  

status/blocked

Cannot start now: waits on an upstream release, another issue, or a recorded revisit trigger

  

status/planned

Triaged with a source-grounded plan and a passed pi review; ready to implement

  

type/bug

Shipped behavior that contradicts the docs or the code's own intent; needs a product code change

  

type/design

An open scope or trade-off question; needs a recorded decision in the owning explanation doc before code

  

type/test-gap

Documented or reviewed behavior that no test pins; usually split from a review finding, needs no product code change

  

type/tracker

Parent issue that tracks child work

No labels

priority/P0

priority/P1

priority/P2

release/v0.1.0

status/blocked

status/planned

type/bug

type/design

type/test-gap

type/tracker

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

natsukium/felis!264

No description provided.