Tune codespell config and fix typos

.pre-commit-config.yaml

@@ -57,11 +57,15 @@ repos:
         exclude: ".envrc$"
 
   - repo: https://github.com/codespell-project/codespell
-    rev: v2.2.4
+    # Configuration lives in pyproject.toml [tool.codespell].
+    rev: v2.4.2
     hooks:
       - id: codespell
-        args: [ "-L", "ue,crate,numer,ser,ratatui,ect,noes,keep-alives", "-S", "Cargo.*,vendor/*" ]
-        exclude: ^docs/developing/doc-standards/tools/vale/styles/
+        additional_dependencies:
+          - tomli; python_version<'3.11'
+        # Mirror central `skip` patterns — pre-commit passes file paths
+        # explicitly, which bypasses codespell's own skip handling.
+        exclude: '^(docs/developing/doc-standards/tools/vale/styles/|vendor/|Cargo\.)'
 
   - repo: https://github.com/rhysd/actionlint
     rev: v1.7.1

docs/developing/doc-standards/canon/doc-types.md

@@ -397,7 +397,7 @@ Optional sections (use when the content warrants):
 
 `docs/developing/decisions/NNNNN-<slug>.md`, where `NNNNN` is zero-padded and `<slug>` is the lowercased, hyphenated form of the decision title (drop articles).
 
-Numbering is monotonic across the project. A new ADR claims the next available number; never re-use a number even for superseded ADRs. The H1 itself doesn't include the number — readers find it via the filename and the directory listing.
+Numbering is monotonic across the project. A new ADR claims the next available number; never reuse a number even for superseded ADRs. The H1 itself doesn't include the number — readers find it via the filename and the directory listing.
 
 ### Voice and length
 

docs/developing/doc-standards/operational/review-checklist.md

@@ -86,7 +86,7 @@ Triggers when the change adds a new ADR or updates the `status` or `date` fields
 *See [`../canon/doc-types.md`](../canon/doc-types.md) § ADR.*
 
 - [ ] New ADR: uses a `status` / `date` header block (not a title/description block).
-- [ ] New ADR: filename follows `NNNNN-<slug>.md` with the next available sequence number; no number is re-used.
+- [ ] New ADR: filename follows `NNNNN-<slug>.md` with the next available sequence number; no number is reused.
 - [ ] Status update on an accepted ADR: only `status` and `date` fields changed — body is untouched.
 - [ ] Change of decision: a new ADR is written (not an edit to the existing body); the prior ADR's `status` is updated to `superseded by ADR-NNNNN` and its body is left unchanged.
 - [ ] Change of decision: the new ADR links to the prior (superseded) ADR in its `More Information` section.

lore-revision/src/nametable.rs

@@ -219,7 +219,7 @@ impl NameTable {
     /// (entries are never removed or relocated), and all callers immediately
     /// copy the result to an owned `String` before the borrow expires.
     pub fn load(&self, hash: u64) -> &str {
-        // TODO(mjansson): Garbage collect nametable by iterating entire merkle tree and rebuiling
+        // TODO(mjansson): Garbage collect nametable by iterating entire merkle tree and rebuilding
         // nametable, cleaning out stale entries
         let data = self.data.read();
         let entry = data.entry_buffer.as_type_slice::<NameTableEntry>();

lore-server/build.rs

@@ -89,7 +89,7 @@ fn protoc_available() -> bool {
         .is_ok_and(|output| output.status.success())
 }
 
-/// Compile the legacy urc.rpc.{Storage,Revision,Repository,Environment}Service protos that now live in this crate. `urc.model` messages stay in `lore-proto`; `extern_path` redirects generated references there so the types are not re-declared.
+/// Compile the legacy urc.rpc.{Storage,Revision,Repository,Environment}Service protos that now live in this crate. `urc.model` messages stay in `lore-proto`; `extern_path` redirects generated references there so the types are not redeclared.
 fn compile_legacy_protos() -> Result<(), Box<dyn Error>> {
     // Declare the codegen inputs unconditionally so Cargo re-runs this script
     // when PROTOC changes or a watched .proto is edited — even on a build where

lore-server/src/settings.rs

@@ -306,7 +306,7 @@ pub struct QuicSettings {
     pub port: i32,
     pub transport_bits_per_second: Option<usize>,
     pub transport_rtt: Option<usize>,
-    /// Keep below a threshold for whatever Load Balancer sits infront of the server
+    /// Keep below a threshold for whatever Load Balancer sits in front of the server
     /// or is expecting responses. If request handlers exceed this reasonable threshold
     /// then assume something has gone wrong and return a timeout response so we can get metrics
     /// and clients don't hang forever

pyproject.toml

@@ -25,3 +25,25 @@ markers = [
     "smoke: marks as smoke test",
     "slow: marks as a slow running test",
 ]
+
+[tool.codespell]
+# Ref: https://github.com/codespell-project/codespell#using-a-config-file
+# Keep options here in sync with the pre-commit `exclude:` regex for codespell —
+# pre-commit passes file paths explicitly, which bypasses these `skip` patterns.
+skip = '.git,.gitignore,.gitattributes,*.svg,vendor,*.lock,*.css,Cargo.*,*/thirdparty/*,*/vale/styles/*'
+check-hidden = true
+# Ignore camelCase identifiers (e.g. test path components like `pathC/subB/`)
+ignore-regex = '\b[a-z]+[A-Z]\w*\b'
+# crate    - Rust keyword (e.g. `crate::`)
+# ratatui  - Rust TUI library name (https://ratatui.rs)
+# ser      - `serde::ser` module / serializer abbreviation
+# noes     - intentional "Oh noes" in test error message
+# keep-alives - HTTP keep-alive (plural form used in some configs)
+# thirdparty - directory name used in path strings (e.g. `native/thirdparty/`)
+# requestor - alt spelling used in lore-transport / system-design docs
+# numer    - numerator abbreviation (preserved from prior inline config)
+# ue, ect  - preserved from prior inline pre-commit config (legacy false positives)
+# collet   - surname (Yann Collet, author of Zstandard) cited in system-design.md
+# accomplis - French loanword in English ("faits accomplis")
+# disjointness - mathematical property of being disjoint (≠ "disjointedness")
+ignore-words-list = 'crate,ratatui,ser,noes,keep-alives,thirdparty,requestor,numer,ue,ect,collet,accomplis,disjointness'

scripts/test/test_conflict.py

@@ -177,7 +177,7 @@ def test_merge_conflict_reset_shows_file_once(new_lore_repo):
 
     # Initial commit with a typo in auto_merged_file
     with repo.open_file(auto_merged_file, "w+") as output_file:
-        output_file.writelines(["conflic in main\n"])
+        output_file.writelines(["conflic in main\n"])  # codespell:ignore conflic
 
     with repo.open_file(conflicting_file, "w+") as output_file:
         output_file.writelines(["Line A\n", "Line B\n", "Line C\n"])
@@ -201,7 +201,7 @@ def test_merge_conflict_reset_shows_file_once(new_lore_repo):
     repo.branch_switch("main")
     repo.branch_create("branch2")
     with repo.open_file(auto_merged_file, "w+") as output_file:
-        output_file.writelines(["conflic in main\n"])
+        output_file.writelines(["conflic in main\n"])  # codespell:ignore conflic
 
     with repo.open_file(conflicting_file, "w+") as output_file:
         output_file.writelines(["Line A\n", "Modified by branch2\n", "Line C\n"])
@@ -218,7 +218,7 @@ def test_merge_conflict_reset_shows_file_once(new_lore_repo):
     # Unstage the auto-merged file (now contains "conflict in main")
     repo.unstage(auto_merged_file)
 
-    # Reset the auto-merged file (reverts to "conflic in main" - branch2's HEAD)
+    # Reset the auto-merged file (reverts to "conflic in main" - branch2's HEAD)  # codespell:ignore conflic
     # BUG: This should show the file once, but it shows twice
     repo.reset(auto_merged_file)