Support for IDENTITY columns

[hab6]

Description

Ingres Connector enhancement providing support for identity columns for use with Ingres databases.

SQLAlchemy Full Test Suite Results

Note: These results exclude several test scripts containing a handful of individual tests that hang.

Configuration:
Ingres 11.2 build 15807 on Windows 10
Python 3.10.7
mock              5.1.0
pluggy            1.4.0
pyodbc            5.1.0
pypyodbc          1.3.6
pytest            8.1.1
SQLAlchemy        2.0.29.dev0
sqlalchemy-ingres 0.0.7.dev1
tox               4.14.2

Result	Total Tests	Passed	Failed	Errors	Skipped	Warnings	Run Time
Before Changes	17568	11737	1316	2765	2077	5	19m 32s
After Changes	17568	14444	485	665	2133	5	21m 26s
Delta	-	2707	(831)	(2100)	56	-	1m 54s

Tests that hang with the code changes in place

The following specific tests of the SQLAlchemy test suite encounter an error without the Ingres connector code changes but now hang with the code changes.

Test	Before Code Changes	After Code Changes
VersioningTest::test_save_update (test\orm\inheritance\test_basic.py)	ERROR: No value specified for mandatory column 'id'	Hang
PartialFlushTest::test_o2m_m2o (test\orm\test_cascade.py)	ERROR: No value specified for mandatory column 'id'	Hang
MapperEventsTest::test_merge (test\orm\test_events.py)	ERROR: No value specified for mandatory column 'id'.	Hang
SessionTransactionTest::test_transactions_isolated (test\orm\test_transaction.py)	ERROR: No value specified for mandatory column 'id'	Hang
AlternateGeneratorTest::test_child_row_switch_two (test\orm\test_versioning.py)	Hang (same result with or without the code changes)	Hang (same result with or without the code changes)

Test scripts that produce worse results with the code changes

The full SQLAlchemy test suite of 17K+ tests was run (excluding the above noted hanging tests) with and without the code changes and results of the ORM related tests were scrutinized. Most ORM test scripts produced better results with the code changes than without the changes. A handful of ORM test scripts produced minor ambiguous results (e.g. fewer errors offset by more skipped tests) and a few test scripts showed very minor worse results with the code changes.

Tests that perform worse with the code changes due to hanging or errors will be dealt with apart from this PR.

Caveat on using IDENTITY columns with X100 tables

Sequences, including identity columns, work for X100 tables only in specific circumstances and are not recommended for X100 tables. See documentation for additional details.

[clach04]

WIP - not yet ready for review, @hab6 working on this at the moment.

[clach04]

👍 lgtm.

Thanks for documenting the SQLAlchemy version.

Caveat on using IDENTITY columns with X100 tables

I recommend warnings be raised if identity columns are used (for DDL only?) and this is an x100 backend. On the fence whether this is a new ticket or something to tackle now (happy to defer).

RE hang/ERROR: No value specified for mandatory column 'id' I think that needs to be the next focus.

I proposed this be squash merged. Any concerns?

[hab6]

Thanks @clach04 for the review and comments.

I will proceed to squash merge this PR and open a couple of new issues for:

• Raise warnings if identity columns are used (for only DDL or also DML?) and this is an x100 backend - Handling sequences and identity columns with X100 tables #44
• Hanging scenarios (whereas prior to these code changes was: ERROR: No value specified for mandatory column 'id') - SQLAlchemy test cases that hang using the Ingres dialect #43

[hab6]

Internal tracking II-13847