Repository navigation
feat: parallel testing infrastructure and SQL Server backend improvements - #2
Merged
Merged
Conversation
- wire the testsuite database pool and pooling providers so scenario tests run in parallel (pytest-xdist -n auto) against per-worker pooled databases instead of the shared scenario database - reset handler skips DROP/CREATE DATABASE when the slot is already clean - enable READ_COMMITTED_SNAPSHOT on pooled databases to avoid deadlock- induced rollback failures in concurrent transaction tests - install activerecord and testsuite from feature/parallel-testing in CI
…ame unification, RICH_AVAILABLE removal, disconnect_async guard)
…onn_args The sqlserver CLI defaults --driver to ODBC Driver 18 for SQL Server, but CI runners only install ODBC Driver 17. The blackbox conn_args did not forward the scenario driver, so query/status/async tests failed with missing ODBC Driver 18. Forward config.driver via --driver so the correct installed driver is used.
…dapter - Fix cli/named_procedure.py adapter: remove invalid get_dialect/execute_query kwargs and ExecutionOptions import (core handle_named_procedure rejects them) - Add named_connections examples (development/production) with 6 connections - Add named_expressions order_expressions (import-safe, prepare_orders_demo helper) - Add named_procedures order_workflow (OrderProcessingProcedure, max_concurrency=1 for pyodbc single-connection concurrency limit) - Add cli demo scripts: named_connection_demo, named_query_demo, named_procedure_demo - Update examples README with new directories
Dialect-level tests (no DB needed) verifying: 1. is_auto_increment=True produces correct auto-increment DDL keyword 2. All standard types (Integer/Text/Boolean/VarChar/Timestamp) compile 3. BooleanType + DEFAULT True is accepted 4. TimestampType maps to correct backend-specific type Prevents recurrence of bugs found in rbac cross-backend testing.
The single uploading job ran coverage run under pytest-xdist without a combine step (59.49% -> 26.46%). Move to pytest-cov reporting and fail on upload errors. Investigation plan recorded locally under .claude/plan/2026-08-26.
Locks the registered-type set (datetime/date/time/Decimal/UUID/dict/list) and documents that primitives intentionally rely on default binding; addresses the backend_mixin gap from the coverage plan.
…rid units 238 cases across the four CLI entrypoints (parser contracts, fake-backend handle() outputs, error exits), 10 adapters bidirectional round-trips, and a SQL Server version grid for pagination/capability/temp-table/hint/MERGE rendering. Defects (XML dict adapter NameError, WKT commas, date-adapter dead branch) recorded in test comments only.
S1: hoist 'import xml.etree.ElementTree as ET' to module scope so SQLServerXMLAdapter dict rendering works instead of raising NameError. S2: _dict_to_wkt now separates coordinate pairs with commas (LINESTRING(1 2, 3 4), POLYGON((0 0, 0 1, 1 1, 0 0))) producing valid WKT. Anchor flip: NameError-pinning XML test replaced with positive round-trip assertions (text/nested/repeated nodes); LineString/Polygon expectations updated to comma-separated WKT.
S4: resolve_connection_config_from_args now reads --encrypt/--no-encrypt and --trust-server-certificate/--no-trust-server-certificate. Flags use a None sentinel default so explicit usage is detectable; explicit flags take precedence over --ssl derivation. The default --ssl auto no longer implies encrypt=True. Default matrix documented in the module docstring. Anchor flip: parser-contract tests assert None sentinel defaults; SSL mapping rewritten as a full five-mode matrix plus flag-combination precedence cases.
S3: SQLServerDateAdapter checks datetime before date (datetime is an isinstance of date) so to_database/from_database truncate datetimes to their date component; the value.date() branches are no longer dead code. Anchor flip: dead-branch pinning test replaced with datetime->date truncation assertions in both directions plus a date-only passthrough case.
S6: query --help epilog now states the three statement input forms (positional argument, -f/--file, stdin pipe) and that no -q short option exists by design. Documentation only; no behaviour change.
Unify placeholder handling per the get_parameter_placeholder() contract: callers provide native '?' markers and _prepare_sql_and_params only performs the SQL Server-specific table hint relocation. Update CRUD tests to use '?'.
Completes the placeholder unification contract: all remaining raw SQL in tests (transaction isolation, connection pool) switches from MySQL-style '%s' to SQL Server native '?' markers.
…en SET statements - Convert all 7 _build_*_sql methods in the SQL Server introspector from f-string identifier interpolation to '?' placeholders, returning (sql, params) tuples consistent with the base IntrospectorMixin contract. - _build_primary_key_sql now returns (sql, params); update sync/async list_columns callers to pass params to executor.execute. - IDENT_CURRENT and DBCC CHECKIDENT now take table/seed as bound parameters instead of raw string embedding. - Add whitelist validation for SET LANGUAGE, SET DATEFORMAT, and SET DEADLOCK_PRIORITY (int range -10..10 or LOW/NORMAL/HIGH).
Test the actual whitelist constants (_SQLSERVER_LANGUAGES / _DATE_FORMATS / _DEADLOCK_PRIORITIES) and fix the bracket-balance assertion to correctly check that every embedded ] is escaped as ]].
… logic - SET DATEFORMAT validation lowercases input, so mdY is valid; keep only genuinely invalid formats in the negative case. - bracket-balance check now actually skips escaped ]] pairs via index skip.
…alidation Transaction manager's set_deadlock_priority now reuses the backend mixin's whitelist (-10..10 or LOW/NORMAL/HIGH) instead of bare f-string concat.
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## release/v1.0.0.dev3 #2 +/- ##
========================================================
+ Coverage 59.49% 72.14% +12.65%
========================================================
Files 85 86 +1
Lines 5742 5783 +41
Branches 927 934 +7
========================================================
+ Hits 3416 4172 +756
+ Misses 2049 1325 -724
- Partials 277 286 +9
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
- cover backend_mixin validation branches (deadlock priority range/ whitelist, language whitelist, dateformat whitelist) and parameterized IDENT_CURRENT/CHECKIDENT queries via mocked cursors - cover introspector parameterized SQL builders (table/column/primary key/index/foreign key/view/trigger) offline - cover __main__.parse_args with custom argv - cover transaction manager set_deadlock_priority backend delegation
|
vistart
added a commit
that referenced
this pull request
Sep 24, 2026
…ents (#2) * feat: parallelize testsuite tests with database pooling for CI - wire the testsuite database pool and pooling providers so scenario tests run in parallel (pytest-xdist -n auto) against per-worker pooled databases instead of the shared scenario database - reset handler skips DROP/CREATE DATABASE when the slot is already clean - enable READ_COMMITTED_SNAPSHOT on pooled databases to avoid deadlock- induced rollback failures in concurrent transaction tests - install activerecord and testsuite from feature/parallel-testing in CI * ci: use -n 8 parallelism for server-type tests * fix(test): check for simultaneous double-hold instead of id uniqueness * feat: CLI async unification, --ssl param; sqlserver fixes (user/username unification, RICH_AVAILABLE removal, disconnect_async guard) * fix(test): pass --driver from scenario config in sqlserver blackbox conn_args The sqlserver CLI defaults --driver to ODBC Driver 18 for SQL Server, but CI runners only install ODBC Driver 17. The blackbox conn_args did not forward the scenario driver, so query/status/async tests failed with missing ODBC Driver 18. Forward config.driver via --driver so the correct installed driver is used. * feat(sqlserver): add named CLI examples and fix named-procedure CLI adapter - Fix cli/named_procedure.py adapter: remove invalid get_dialect/execute_query kwargs and ExecutionOptions import (core handle_named_procedure rejects them) - Add named_connections examples (development/production) with 6 connections - Add named_expressions order_expressions (import-safe, prepare_orders_demo helper) - Add named_procedures order_workflow (OrderProcessingProcedure, max_concurrency=1 for pyodbc single-connection concurrency limit) - Add cli demo scripts: named_connection_demo, named_query_demo, named_procedure_demo - Update examples README with new directories * test(expression): add serialization round-trip coverage for SQL Server expressions * test: add is_auto_increment/DDL regression tests Dialect-level tests (no DB needed) verifying: 1. is_auto_increment=True produces correct auto-increment DDL keyword 2. All standard types (Integer/Text/Boolean/VarChar/Timestamp) compile 3. BooleanType + DEFAULT True is accepted 4. TimestampType maps to correct backend-specific type Prevents recurrence of bugs found in rbac cross-backend testing. * feat(dialect): declare supports_schema; add cross-schema provider fixtures * test(provider): provision mixed default/non-default schema fixtures * ci(test): merge xdist coverage via pytest-cov; restore codecov fidelity The single uploading job ran coverage run under pytest-xdist without a combine step (59.49% -> 26.46%). Move to pytest-cov reporting and fail on upload errors. Investigation plan recorded locally under .claude/plan/2026-08-26. * test(backend): adapter suggestion cache behavior without a live server Locks the registered-type set (datetime/date/time/Decimal/UUID/dict/list) and documents that primitives intentionally rely on default binding; addresses the backend_mixin gap from the coverage plan. * test(sqlserver): CLI black-box suites, adapters and dialect version-grid units 238 cases across the four CLI entrypoints (parser contracts, fake-backend handle() outputs, error exits), 10 adapters bidirectional round-trips, and a SQL Server version grid for pagination/capability/temp-table/hint/MERGE rendering. Defects (XML dict adapter NameError, WKT commas, date-adapter dead branch) recorded in test comments only. * fix(adapters): module-level ET import and WKT point separators S1: hoist 'import xml.etree.ElementTree as ET' to module scope so SQLServerXMLAdapter dict rendering works instead of raising NameError. S2: _dict_to_wkt now separates coordinate pairs with commas (LINESTRING(1 2, 3 4), POLYGON((0 0, 0 1, 1 1, 0 0))) producing valid WKT. Anchor flip: NameError-pinning XML test replaced with positive round-trip assertions (text/nested/repeated nodes); LineString/Polygon expectations updated to comma-separated WKT. * fix(cli): honour encrypt/trust flags with defined precedence over --ssl S4: resolve_connection_config_from_args now reads --encrypt/--no-encrypt and --trust-server-certificate/--no-trust-server-certificate. Flags use a None sentinel default so explicit usage is detectable; explicit flags take precedence over --ssl derivation. The default --ssl auto no longer implies encrypt=True. Default matrix documented in the module docstring. Anchor flip: parser-contract tests assert None sentinel defaults; SSL mapping rewritten as a full five-mode matrix plus flag-combination precedence cases. * fix(adapters): date adapter branch ordering S3: SQLServerDateAdapter checks datetime before date (datetime is an isinstance of date) so to_database/from_database truncate datetimes to their date component; the value.date() branches are no longer dead code. Anchor flip: dead-branch pinning test replaced with datetime->date truncation assertions in both directions plus a date-only passthrough case. * docs(cli): clarify statement input forms S6: query --help epilog now states the three statement input forms (positional argument, -f/--file, stdin pipe) and that no -q short option exists by design. Documentation only; no behaviour change. * fix(backend): remove %s->? placeholder guessing at execute layer Unify placeholder handling per the get_parameter_placeholder() contract: callers provide native '?' markers and _prepare_sql_and_params only performs the SQL Server-specific table hint relocation. Update CRUD tests to use '?'. * test: use native '?' placeholders in remaining SQL Server tests Completes the placeholder unification contract: all remaining raw SQL in tests (transaction isolation, connection pool) switches from MySQL-style '%s' to SQL Server native '?' markers. * fix(introspection): parameterize SQL Server introspector SQL and harden SET statements - Convert all 7 _build_*_sql methods in the SQL Server introspector from f-string identifier interpolation to '?' placeholders, returning (sql, params) tuples consistent with the base IntrospectorMixin contract. - _build_primary_key_sql now returns (sql, params); update sync/async list_columns callers to pass params to executor.execute. - IDENT_CURRENT and DBCC CHECKIDENT now take table/seed as bound parameters instead of raw string embedding. - Add whitelist validation for SET LANGUAGE, SET DATEFORMAT, and SET DEADLOCK_PRIORITY (int range -10..10 or LOW/NORMAL/HIGH). * test: add SET whitelist and identifier quoting balance tests * test: fix SQL Server whitelist and bracket-balance security tests Test the actual whitelist constants (_SQLSERVER_LANGUAGES / _DATE_FORMATS / _DEADLOCK_PRIORITIES) and fix the bracket-balance assertion to correctly check that every embedded ] is escaped as ]]. * test: fix SQL Server security test dateformat casing and bracket-skip logic - SET DATEFORMAT validation lowercases input, so mdY is valid; keep only genuinely invalid formats in the negative case. - bracket-balance check now actually skips escaped ]] pairs via index skip. * fix(sqlserver): delegate set_deadlock_priority to backend whitelist validation Transaction manager's set_deadlock_priority now reuses the backend mixin's whitelist (-10..10 or LOW/NORMAL/HIGH) instead of bare f-string concat. * ci: update testsuite branch to release/v1.0.0.dev17 * ci: update rhosocial-activerecord branch to release/v1.0.0.dev30 * ci: update testsuite branch to release/v1.0.0.dev17 * test: add targeted coverage for low-coverage patch areas - cover backend_mixin validation branches (deadlock priority range/ whitelist, language whitelist, dateformat whitelist) and parameterized IDENT_CURRENT/CHECKIDENT queries via mocked cursors - cover introspector parameterized SQL builders (table/column/primary key/index/foreign key/view/trigger) offline - cover __main__.parse_args with custom argv - cover transaction manager set_deadlock_priority backend delegation * chore: add changelog fragment for PR #2
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
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.
Description
This PR brings parallel testing support and coordinated SQL Server backend changes to the
release/v1.0.0.dev3line, aligned withpython-activerecordPR #118 (feature/parallel-testingmerged intorelease/v1.0.0.dev30).Issues: (add issue numbers here)
Type of change
feat: A new featuredocs: Documentation only changes (if related to the new feature)test: Adding missing tests or correcting existing tests (for the new feature)chore: Changes to the build process or auxiliary tools and libraries such as documentation generation (if related to the new feature)ci: Changes to our CI configuration files and scripts (if related to the new feature)build: Changes that affect the build system or external dependencies (if related to the new feature)revert: Reverts a previous commitBreaking Change
Related Repositories/Packages
python-activerecord/release/v1.0.0.dev30— parallel testing infrastructure, expression/serialization, CLI unification (PR #118)python-activerecord-testsuite/release/v1.0.0.dev17— parallel-testing plugin, pool/serial mechanism, two-level execution classification, cross-schema fixtures, shared expression round-trip utils (PR #20)Changes
CI / Infrastructure
-n 8parallelism for server-type testsrhosocial-activerecordbranch torelease/v1.0.0.dev30rhosocial-activerecord-testsuitebranch torelease/v1.0.0.dev17SQL Server-Specific Fixes
set_deadlock_priorityto backend whitelist validation%s→?placeholder guessing at execute layer--sslSQL Server-Specific Features
supports_schema; add cross-schema provider fixtures?placeholders in remaining SQL Server testsSecurity Hardening
CLI
--sslparam; sqlserver fixes (user/username unification, RICH_AVAILABLE removal, disconnect_async guard)--driverfrom scenario config in sqlserver blackbox conn_argsAdapter / Test Enhancements
is_auto_increment/DDL regression testsTesting
Test Plan:
(Skipped local verification; CI will run full SQL Server matrix after PR creation.)
Checklist
README.md, etc.).changelog.d/<issue_number>.<type>.md).Additional Notes
Targets release:
release/v1.0.0.dev3