ci: add the typos spell check (#632)

Author: azimafroozeh

.github/workflows/ci.yml

@@ -93,6 +93,7 @@ jobs:
   check_agents_md:
     name: Check AGENTS.md Links
     runs-on: ubuntu-latest
+    timeout-minutes: 10
     permissions:
       contents: read
     steps:
@@ -133,6 +134,18 @@ jobs:
       - name: Verify documentation structure and RFC registry
         run: python3 scripts/check-docs.py
 
+      # Spelling gate: docs/dev/ci.md, "Pull-request gates". typos also reads
+      # a sibling typos.toml / _typos.toml, so refuse a second config first.
+      - name: Refuse a second typos config
+        run: |
+          if [ -e typos.toml ] || [ -e _typos.toml ]; then
+            echo "::error::every typos exemption lives in .typos.toml; remove typos.toml / _typos.toml"
+            exit 1
+          fi
+
+      - name: Check spelling
+        uses: crate-ci/typos@d43b6c087ac471e2ea7b8af622ff15f05c0c365b # v1.50.1
+
   workflow_action_pins:
     name: Check Workflow Action Pins
     runs-on: ubuntu-latest

.typos.toml

@@ -0,0 +1,37 @@
+# Known-misspelling check (crate-ci/typos). CI runs it from the repository
+# root, and so must a local run. Every exemption lives in this file, one
+# comment per entry saying why the token is intended where it occurs.
+
+[files]
+ignore-hidden = false # workflows, issue templates and .cargo/ hold prose too
+extend-exclude = [
+  ".git/",
+  ".typos.toml", # names the tokens it exempts
+]
+
+[default]
+# The "mis-" prefix (mis-renamed, mis-estimate); a bare "mis" still fires.
+extend-ignore-re = ['\bmis-\w+']
+
+# Tokens that occur in prose and code alike.
+[default.extend-words]
+AKS = "AKS"                 # Azure Kubernetes Service (RFC 0029 assets)
+BA = "BA"                   # AB/BA measurement-order labels (RFCs 0023, 0031); also the hex fragment of the short commit SHAs 2aab48ba (RFC 0019) and da466ba (vocabulary guard test). Cost: a bare "ba" typed for "by" passes
+fof = "fof"                 # friend-of-friend: pattern variable in test queries and the query reference
+unparseable = "unparseable" # accepted variant; appears in operator-facing error strings and docs
+
+# Tokens that exist only inside Rust sources; the same misspelling in
+# Markdown still fires.
+[type.rust.extend-identifiers]
+write_applyable_state = "write_applyable_state"                 # cluster test helper: state that can be applied
+seed_applyable_state = "seed_applyable_state"                   # same family
+write_cluster_applyable_state = "write_cluster_applyable_state" # CLI test helper, same meaning
+
+[type.rust.extend-words]
+alph = "alph"         # prefix-filter probe value in literal_filters.rs
+celle = "celle"       # matrix cell "e" in server/tests/multi_graph.rs (cella … cellj)
+froms = "froms"       # plural of `from`, paired with `tos`, in the DST edge fixtures
+nmae = "nmae"         # deliberately wrong parameter name (stored_queries.rs, vocabulary guard lib.rs tests)
+pn = "pn"             # external-id prefix ("pn:12345") in a consistency fixture
+propert = "propert"   # deliberate substring assertion on a typecheck error message
+referenc = "referenc" # deliberate substring match on Lance's error text (table_ops.rs, branch_control.rs)

AGENTS.md

@@ -136,6 +136,7 @@ cargo clippy --workspace --all-targets --locked -- \
 bash scripts/check-agents-md.sh
 python3 scripts/check-docs.py
 python3 scripts/check-workflow-action-pins.py
+typos                                   # from the repository root; version pinned in ci.yml; exemptions in .typos.toml
 ```
 
 S3 suites require `OMNIGRAPH_S3_TEST_BUCKET` and the documented `AWS_*`

crates/omnigraph-dst/tests/scenarios.rs

@@ -3741,7 +3741,7 @@ fn dst_merge_duplicates_born_on_both_edge() {
         omnigraph::dst_clock::install_logical_clock();
         let storage: Arc<dyn StorageAdapter> = Arc::new(ObjectStorageAdapter::in_memory());
         let db = Omnigraph::init_with_storage(
-            "shared-memory://dst-classe-probe",
+            "shared-memory://dst-class-probe",
             TEST_SCHEMA,
             storage.clone(),
             InitOptions::default(),

crates/omnigraph-server/tests/data_routes.rs

@@ -2631,7 +2631,7 @@ async fn change_concurrent_updates_same_key_return_typed_pre_effect_conflicts()
         .unwrap();
     let app = build_app(state);
 
-    // Spawn N=8 concurrent UPDATEs on Alice (from test.jsonl, age=30 at V0)
+    // Spawn N=8 concurrent UPDATE mutations on Alice (from test.jsonl, age=30 at V0)
     // writing distinct ages.
     const N: usize = 8;
     let mut handles = Vec::with_capacity(N);

crates/omnigraph/src/db/manifest/tests.rs

@@ -1401,7 +1401,7 @@ async fn test_batch_create_table_versions_allows_owner_branch_handoff_at_same_ve
 }
 
 /// Regression (PR #307 review — Cursor Bugbot High + Codex P2): the post-publish
-/// fold (`#1b`) must reflect an owner-branch handoff. A handoff UPDATEs a
+/// fold (`#1b`) must reflect an owner-branch handoff. A handoff updates a
 /// `table_version` row IN PLACE at the SAME Lance version with a new
 /// `table_branch` — merge-insert `UpdateAll` on the deterministic
 /// `version_object_id(table_key, version)`, so `__manifest` ends with one row

crates/omnigraph/src/exec/query.rs

@@ -2871,7 +2871,7 @@ async fn execute_expand_bound(
                 .downcast_ref::<StringArray>()
                 .ok_or_else(|| OmniError::manifest(format!("edge '{}' is not Utf8", key_col)))?
                 .clone();
-            let opps = batch
+            let opposites = batch
                 .column_by_name(opp_col)
                 .ok_or_else(|| OmniError::manifest(format!("edge batch missing '{}'", opp_col)))?
                 .as_any()
@@ -2888,7 +2888,7 @@ async fn execute_expand_bound(
             for r in 0..batch.num_rows() {
                 // Undirected probes both orientations; a self-loop row would
                 // match the same wide row through both, so emit it only once.
-                if probe_idx == 1 && keys.value(r) == opps.value(r) {
+                if probe_idx == 1 && keys.value(r) == opposites.value(r) {
                     continue;
                 }
                 let Some(wide_rows) = rows_by_src.get(keys.value(r)) else {
@@ -2897,7 +2897,7 @@ async fn execute_expand_bound(
                 for &wide_row in wide_rows {
                     matches.push((
                         wide_row,
-                        opps.value(r).to_string(),
+                        opposites.value(r).to_string(),
                         batch_idx,
                         r,
                         edge_ids.value(r).to_string(),
@@ -3296,7 +3296,7 @@ async fn execute_expand_bfs(
                         .ok_or_else(|| {
                             OmniError::manifest(format!("edge '{}' is not Utf8", key_col))
                         })?;
-                    let opps = batch
+                    let opposites = batch
                         .column_by_name(opp_col)
                         .ok_or_else(|| {
                             OmniError::manifest(format!("edge batch missing '{}'", opp_col))
@@ -3308,7 +3308,7 @@ async fn execute_expand_bfs(
                         })?;
                     for r in 0..batch.num_rows() {
                         let k = src.interner.get_or_insert(keys.value(r));
-                        let o = src.interner.get_or_insert(opps.value(r));
+                        let o = src.interner.get_or_insert(opposites.value(r));
                         src.neighbor_map.entry(k).or_default().push(o);
                     }
                 }

crates/omnigraph/tests/failpoints.rs

@@ -5394,7 +5394,7 @@ async fn s3_load_recovers_after_publisher_failure_without_reopen() {
     }
 
     // Same-handle follow-up load: the entry heal LISTs __recovery/ on
-    // S3, rolls the sidecar forward, DELETEs it, and the write lands.
+    // S3, rolls the sidecar forward, issues a DELETE for it, and the write lands.
     load_jsonl(
         &db,
         r#"{"type":"Person","data":{"name":"Bob","age":25}}

crates/omnigraph/tests/writes.rs

@@ -923,7 +923,7 @@ async fn overlapping_delete_predicates_do_not_double_count_affected() {
 }
 
 /// The overlap-exclusion filter must use SQL `IS NOT TRUE`, not `NOT`: a prior
-/// delete predicate referencing a NULLable column must NOT drop a later
+/// delete predicate referencing a nullable column must NOT drop a later
 /// statement's matching row just because that column is NULL (SQL UNKNOWN).
 /// With `NOT (age > 30)`, a row with NULL `age` makes the clause UNKNOWN and the
 /// row is filtered out of `deleted_ids` — skipping its cascade (orphaned edges),

docs/dev/ci.md

@@ -58,7 +58,22 @@ the same context also rejects any pull request whose own diff adds a
 conflict-marker line in any file type, annotating each offending file and
 line; markers already on the base branch never fail an unrelated pull
 request. There is no exemption; a document that must quote a conflict block
-indents the markers one space.
+indents the markers one space. After the documentation checks, the same
+context runs `typos` (`crate-ci/typos`, pinned by commit) over every tracked
+text file, hidden paths such as `.github/` included; the tool itself skips
+`Cargo.toml` manifests, lock files and binaries. It matches each word against
+a list of known misspellings, not a dictionary: an unknown word never fires,
+and an identifier fires only when one of the words it splits into is on the
+list (a CamelCase fragment or a short abbreviation can be one). A
+flagged token that is correct where it occurs gets one commented line in
+`.typos.toml`: under `[default.extend-words]` when it appears in prose and
+code alike, under `[type.rust.extend-words]` (or `extend-identifiers` for one
+exact identifier) when it exists only in Rust sources, so the same
+misspelling in Markdown still fires; a hyphenated prefix goes in
+`extend-ignore-re`; a generated text file gets an `extend-exclude` glob.
+Every exemption lives in that one file: the job refuses a sibling
+`typos.toml` or `_typos.toml`, and CI ignores a config file in a
+subdirectory (a local run inside that subdirectory would not).
 
 `Graph Vocabulary Guard` remains a required reporting context, but its
 substrate-sized audit steps are currently disabled everywhere (decision of
@@ -191,11 +206,12 @@ python3 scripts/check-workflow-action-pins.py
 python3 scripts/check-release-vocabulary-gates.py
 python3 scripts/check-container-binary-contract.py
 python3 scripts/check-azure-admission-boundary.py
+typos                       # from the repository root; a subdirectory run scans only that subtree
 actionlint .github/workflows/*.yml
 shellcheck scripts/*.sh
 ```
 
-`actionlint` and `shellcheck` are developer tools, not installed by Cargo. Run the applicable subset when a change does not touch their surface.
+`typos` (`cargo install typos-cli --locked --version 1.50.1`, the version `ci.yml` pins; the misspelling list grows per release, so a newer local binary can flag words CI accepts), `actionlint` and `shellcheck` are developer tools, not workspace dependencies. Run the applicable subset when a change does not touch their surface.
 
 ## Release workflows