Implement temporal comparisons

[dbussink]

This is currently missing and leads to incorrect types to be returned for LEAST & GREATEST as comparison functions.

There's a little mismatch here in behavior compared to MySQL which I argue is actually a bug in MySQL. In MySQL, a temporal type always has the binary collation:

mysql> select NOW(6), collation(NOW(6));
+----------------------------+-------------------+
| NOW(6)                     | collation(NOW(6)) |
+----------------------------+-------------------+
| 2025-02-19 15:33:21.732301 | binary            |
+----------------------------+-------------------+
1 row in set (0.00 sec)

On MySQL 8.4, this results in:

mysql> select GREATEST(NOW(6), NOW(6)), collation(GREATEST(NOW(6), NOW(6)));
+----------------------------+-------------------------------------+
| GREATEST(NOW(6), NOW(6))   | collation(GREATEST(NOW(6), NOW(6))) |
+----------------------------+-------------------------------------+
| 2025-02-19 15:35:00.921308 | latin1_swedish_ci                   |
+----------------------------+-------------------------------------+
1 row in set (0.00 sec)

But on MySQL 8.0, it returns:

mysql> select GREATEST(NOW(6), NOW(6)), collation(GREATEST(NOW(6), NOW(6)));
+----------------------------+-------------------------------------+
| GREATEST(NOW(6), NOW(6))   | collation(GREATEST(NOW(6), NOW(6))) |
+----------------------------+-------------------------------------+
| 2025-02-19 15:35:00.921308 | utf8mb4_0900_ai_ci                  |
+----------------------------+-------------------------------------+
1 row in set (0.00 sec)

Neither of these collations make sense, because it really should not change the collation and return binary still. That is what Vitess still does with the changes here (hence the addition to the test framework to allow skipping the collation check).

I'll also report the issue upstream to make it behave correctly there as well.

Marked this also for backporting since using these types today leads to a panic on queries.

Related Issue(s)

Fixes #17750

Checklist

• "Backport to:" labels have been added if this change should be back-ported to release branches
• If this change is to be back-ported to previous releases, a justification is included in the PR description
• Tests were added or are not required
• Did the new or modified tests pass consistently locally and on CI?
• Documentation was added or is not required

[vitess-bot]

Review Checklist

Hello reviewers! 👋 Please follow this checklist when reviewing this Pull Request.

General

• Ensure that the Pull Request has a descriptive title.
• Ensure there is a link to an issue (except for internal cleanup and flaky test fixes), new features should have an RFC that documents use cases and test cases.

Tests

• Bug fixes should have at least one unit or end-to-end test, enhancement and new features should have a sufficient number of tests.

Documentation

• Apply the release notes (needs details) label if users need to know about this change.
• New features should be documented.
• There should be some code comments as to why things are implemented the way they are.
• There should be a comment at the top of each new or modified test to explain what the test does.

New flags

• Is this flag really necessary?
• Flag names must be clear and intuitive, use dashes (-), and have a clear help text.

If a workflow is added or modified:

• Each item in Jobs should be named in order to mark it as required.
• If the workflow needs to be marked as required, the maintainer team must be notified.

Backward compatibility

• Protobuf changes should be wire-compatible.
• Changes to _vt tables and RPCs need to be backward compatible.
• RPC changes should be compatible with vitess-operator
• If a flag is removed, then it should also be removed from vitess-operator and arewefastyet, if used there.
• vtctl command output order should be stable and awk-able.

[codecov]

Codecov Report

Attention: Patch coverage is 80.55191% with 148 lines in your changes missing coverage. Please review.

Project coverage is 67.42%. Comparing base (2118bc3) to head (4171a5e).

Files with missing lines	Patch %	Lines
go/vt/vtgate/evalengine/eval_temporal.go	17.72%	65 Missing ⚠️
go/vt/vtgate/evalengine/fn_compare.go	89.03%	25 Missing ⚠️
go/vt/vtgate/evalengine/compiler_asm_push.go	0.00%	21 Missing ⚠️
go/vt/vtgate/evalengine/compiler_asm.go	60.97%	16 Missing ⚠️
go/vt/vtgate/evalengine/compiler.go	44.44%	15 Missing ⚠️
go/vt/vtgate/evalengine/expr_bvar.go	0.00%	3 Missing ⚠️
go/vt/vtgate/evalengine/expr_column.go	0.00%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #17826      +/-   ##
==========================================
- Coverage   67.45%   67.42%   -0.03%     
==========================================
  Files        1592     1592              
  Lines      258167   258554     +387     
==========================================
+ Hits       174143   174338     +195     
- Misses      84024    84216     +192     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[systay]

shouldn't we use skipCollationCheck here?

[dbussink]

@systay This currently doesn't do collations checks anyway, so I don't think it really matters.

[dbussink]

@systay This did remind me that we actually should validate the collation. We still ignore this flag though, since our evalengine and compiler should always match.

The ignore collation flag is for where explicitly don't 100% match MySQL (due to bugs, arguably broken behavior etc).

[systay]

nitpick: another solution would be to use the variable name datetimes in the functions, and keep import named datetime

[systay]

LGTM

[dbussink]

skipCollationCheck is needed because of the issue in MySQL described in mysql/mysql-server#602.