release 0.7.0

.changes/0.7.0.md

@@ -0,0 +1,3 @@
+## 0.7.0 - 2026-07-13
+### Added
+* BigQuery platform adapter: run SQL evals and data checks on BigQuery. Install `evaldata[bigquery]`.

CHANGELOG.md

@@ -6,6 +6,10 @@ adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html),
 and is generated by [Changie](https://github.com/miniscruff/changie).
 
 
+## 0.7.0 - 2026-07-13
+### Added
+* BigQuery platform adapter: run SQL evals and data checks on BigQuery. Install `evaldata[bigquery]`.
+
 ## 0.6.0 - 2026-07-09
 ### Added
 * Snowflake Cortex Analyst solver: evaluate a deployed Cortex Analyst through the usual scoring cascade. `CortexAnalystSolver` sends a natural-language question to the Cortex Analyst REST endpoint (reusing your Snowflake connection's session token) and returns the SQL it generates, which evaldata runs and scores against a gold query. Install the `cortex` extra.

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "evaldata"
-version = "0.6.0"
+version = "0.7.0"
 description = "Evaluate AI-generated SQL and text-to-SQL systems with pytest."
 readme = "README.md"
 license = "Apache-2.0"

tests/platforms/conftest.py

@@ -170,7 +170,7 @@ class BigQueryFixtures:
 
     BigQuery's driver reports precise result-column types directly, so `reports_types` is `True`
     with no probe needed. `UNNEST` supplies the row-list form GoogleSQL lacks as `VALUES ... AS
-    t(n)`; the `slow_query` counts a large generated array, which scans ~0 bytes yet runs long
+    t(n)`; the `slow_query` aggregates a large generated array, which scans ~0 bytes yet runs long
     enough to overrun a sub-second budget.
     """
 
@@ -183,7 +183,7 @@ class BigQueryFixtures:
     duplicate_column_names: str = "SELECT 1 AS x, 2 AS x"
     references_missing_table: str = "SELECT * FROM does_not_exist_xyz"
     parse_error: str = "SLECT 1"
-    slow_query: str = "SELECT COUNT(*) AS n FROM UNNEST(GENERATE_ARRAY(1, 100000000))"
+    slow_query: str = "SELECT SUM(n) AS n FROM UNNEST(GENERATE_ARRAY(1, 10000000)) AS n"
     reports_types: bool = True
     renames_duplicate_columns: bool = True