Codestin Search App

bschoening · 2026-05-26T19:35:30Z

A recreated version of #1290 (which was closed accidentally)

absurdfarce · 2026-05-29T17:25:57Z

+                     "to hang indefinitely. If you want to use the Eventlet reactor, you "
+                     "need to install the `futurist` package to allow the driver to use "
+                     "the GreenThreadPoolExecutor. See https://github.com/eventlet/eventlet/issues/508 "
+                     "for more details."))


So I think this was the only point left from our conversation on the prior PR @bschoening. Based on my earlier comment I'm inclined to say we should just remove this check (and executor class set operation) completely.

Whaddya think?

…ars to have fixed the issue on their side. Enables removal of futurist dependency in it's entirety.

absurdfarce · 2026-06-12T16:12:15Z

        with self.VerifiedFunction(self, **kwargs) as vf:
            fn_meta = self.keyspace_function_meta[vf.signature]
-            self.assertRegex(fn_meta.as_cql_query(), "CREATE FUNCTION.*\) RETURNS NULL ON NULL INPUT RETURNS .*")
+            self.assertRegex(fn_meta.as_cql_query(), "CREATE FUNCTION.*) RETURNS NULL ON NULL INPUT RETURNS .*")


I'm not trying to remember where this change (and the similar change above) came from. These tests are failing on Jenkins due to these changes... and they're failing when I run the test locally as well.

I think I made these changes and I think I convinced myself that they were okay via local runs... but something went wrong somewhere along the way.

Oh, I think I get it... they just need to be raw strings. Working on fixing this now.

absurdfarce · 2026-06-16T16:22:04Z

Jenkins run for this appears clean (or at least clean of any regressions). @bschoening are you good with this one as it stands? I think this is the last thing we wanted to get in for 3.30.1.

Copilot

Pull request overview

This PR removes legacy Python-version conditionals and drops the futurist dependency, simplifying both runtime code and tests now that the driver targets modern Python versions.

Changes:

Remove futurist from test dependencies and delete related eventlet-specific executor selection logic.
Simplify Cluster initialization to always use concurrent.futures.ThreadPoolExecutor.
Remove Python-version-specific branches and tighten up formatting in multiple unit/integration tests.

Reviewed changes

Copilot reviewed 11 out of 11 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
`tox.ini`	Removes `futurist` from tox test dependencies.
`test-requirements.txt`	Drops `futurist` from test requirements.
`cassandra/cluster.py`	Removes `_create_thread_pool_executor` and uses `ThreadPoolExecutor` directly.
`tests/__init__.py`	Removes `futurist`-based executor selection for eventlet tests.
`tests/unit/test_row_factories.py`	Removes old Python-version workaround; updates/loosens namedtuple regression test.
`tests/unit/test_policies.py`	Minor formatting cleanup in assertions.
`tests/unit/advanced/test_insights.py`	Removes Python-version conditional namespace handling and unused imports.
`tests/integration/standard/test_metadata.py`	Removes old skips; updates regex strings to raw literals.
`tests/integration/cqlengine/columns/test_validation.py`	Removes obsolete Python-version branches and applies formatting cleanups.
`tests/integration/cqlengine/base.py`	Removes Python-version conditional aliasing and unused import.
`tests/integration/advanced/graph/fluent/__init__.py`	Removes Python 2 `long` type handling.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

absurdfarce · 2026-06-16T17:33:57Z


-        if PROTOCOL_VERSION < 3:
-            raise unittest.SkipTest(
-                "Protocol 3.0+ is required for UDT change events, currently testing against %r"
-                % (PROTOCOL_VERSION,))
-
-        if sys.version_info[0:2] != (2, 7):
-            raise unittest.SkipTest('This test compares static strings generated from dict items, which may change orders. Test with 2.7.')
-
        cluster = TestCluster()
        session = cluster.connect()



This is kinda the point. We no longer test with anything less than 3.x so having this constraint makes no sense.

absurdfarce · 2026-06-16T23:26:50Z

-
-        for r in rows:
-            self.assertEqual(r.col0, self.long_rows[0][0])
+        named_tuple_factory(self.long_colnames, self.long_rows)


I disagree that this is a no-op as it stands. The underlying issue with PYTHON-893 was that namedtuple() would throw an exception if it were passed more than 255 parameters. unittest will interpret an unhandled exception from a test case as a failure of that test case so just making the call (with appropriately contrived arguments) should be enough to trigger the error if it occurs.

Now, I'm perfectly willing to concede that the error shouldn't occur for versions of Python we're actually testing against since this issue has been resolved in the Python runtime (as referenced in the comments). So what we're guarding against here is a regression in how the Python runtime handles namedtuple() with lots of args. If we decide we aren't worried about that we can absolutely remove this test... but if we are I don't think we need to go through the extra steps suggested by Copilot.

Since this Cython limitation is now explicitly tested upstream (per Python issue18896) in test_collections.py, we probably don't need to duplicate it here. It might just become technical debt/clutter down the road. If you feel strongly about keeping it, though, I'm okay with it!

diff -r 79d52e7d472d -r 7454ca88aacb Lib/test/test_collections.py --- a/Lib/test/test_collections.py Fri Dec 16 19:06:51 2016 +0200 +++ b/Lib/test/test_collections.py Fri Dec 16 19:19:02 2016 +0200 @@ -319,8 +319,7 @@ self.assertEqual(Dot(1)._replace(d=999), (999,)) self.assertEqual(Dot(1)._fields, ('d',)) - # n = 5000 - n = 254 # SyntaxError: more than 255 arguments: + n = 5000 names = list(set(''.join([choice(string.ascii_letters) for j in range(10)]) for i in range(n)))

An entirely reasonable point.

I also just realized that we're already doing something different in this PR! The rationale for removing the warning about the thread pool impl when using eventlet was that eventlet had already fixed the underlying issue so new versions of the driver wouldn't need to worry about older versions. It's a pretty straightforward parallel; if we do it there we probably should do the same in this case.

I have a slight concern for folks upgrading existing applications which may be working off of a dated set of dependencies... that could potentially undermine the approach I argued for with eventlet. But it's going away in the next release anyway (and isn't used much now) so... I'm not super-worried about it.

Regardless, I think we're in agreement on this test; it can just be removed. I'm also going to remove the check logic that was put in place as part of PYTHON-893.

bschoening · 2026-06-16T20:27:22Z

-
-NAMEDTUPLE_CREATION_BUG = sys.version_info >= (3,) and sys.version_info < (3, 7)
-
 class TestNamedTupleFactory(TestCase):


Based on the Co-pilot comment, it seems like the whole test function can be dropped now (post 3.7)

Discussed below. Fundamentally this comes down to whether we want to test for regressions or not.

bschoening · 2026-06-16T20:28:00Z

-            self.assertEqual(r.col0, self.long_rows[0][0])
+        named_tuple_factory(self.long_colnames, self.long_rows)

    def test_creation_no_warning_on_short_column_list(self):


Can this be dropped also?

This is actually testing a different case; we're validating proper functionality for a shorter (arguably "normal") length column list. I'd argue for something similar to what I said above; if we decide we aren't worried about regressions in the handling of namedtuple args we can delete this file in it's entirety. I'm not arguing for or against it necessarily... I can see arguments both ways.

absurdfarce · 2026-06-16T23:40:05Z

Mentioned to @bschoening in other communication but I'll make a note of it here too; if we remove the test cases above (because Python runtimes we support no longer have this issue) we should probably remove the workaround added for PYTHON-893 as well.

bschoening and others added 3 commits April 29, 2026 14:02

CASSPYTHON-18: removed obsolete check for python version > 3.7+

7cdb996

Fix for test_row_factories.py

58d2566

Fix for test_metadata.py

8b5ad09

absurdfarce reviewed May 29, 2026

View reviewed changes

Removing special handling for evenlet connections since eventlet appe…

b3c45be

…ars to have fixed the issue on their side. Enables removal of futurist dependency in it's entirety.

absurdfarce reviewed Jun 12, 2026

View reviewed changes

Fix test_metadata correctly

a5aafb2

absurdfarce self-requested a review June 12, 2026 20:26

Merge branch 'trunk' into CASSPYTHON-18

fe6f38f

absurdfarce requested a review from Copilot June 16, 2026 17:17

Copilot started reviewing on behalf of absurdfarce June 16, 2026 17:18 View session

Copilot AI reviewed Jun 16, 2026

View reviewed changes

bschoening commented Jun 16, 2026

View reviewed changes

Remove unnecessary import

eee7523


		NAMEDTUPLE_CREATION_BUG = sys.version_info >= (3,) and sys.version_info < (3, 7)

		class TestNamedTupleFactory(TestCase):

Conversation

bschoening commented May 26, 2026 • edited by absurdfarce Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

absurdfarce commented Jun 16, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

absurdfarce Jun 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

bschoening Jun 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

absurdfarce commented Jun 16, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

bschoening commented May 26, 2026 •

edited by absurdfarce

Loading

absurdfarce Jun 16, 2026 •

edited

Loading

bschoening Jun 16, 2026 •

edited

Loading