apache / apache/datafusion

Run sqllogictests multiple times with random fuzzed configurations

Open
#9,746 1 comment 1 reaction 0 assignees View on GitHub
enhancement
Dominant language
Rust
Stars
9.3k
Forks
2.4k
Avg merge
3d 7h
Merged PRs (30d)
344

Description

### Is your feature request related to a problem or challenge?

A thought coming off of fixing #9736 is that we could likely proactively identify issues like this by modifying sqllogictests to run multiple times with a variety of randomized configs.

### Describe the solution you'd like

Perhaps this function https://github.com/apache/arrow-datafusion/blob/6c6305159711fbca43ff7798faa52aaebcb2e7d3/datafusion/sqllogictest/src/test_context.rs#L67
could be modified to set random configs over a range of valid values.

Then this could be modified to run in a loop a set number of times based on a passed option with a default.

https://github.com/apache/arrow-datafusion/blob/6c6305159711fbca43ff7798faa52aaebcb2e7d3/datafusion/sqllogictest/bin/sqllogictests.rs#L90

### Describe alternatives you've considered

If this results in tests taking too long, the default could be 1 run, but occasionally (prior to a release perhaps) a longer test with many runs could be run.

### Additional context

_No response_

Contributor guide

Open the contributing guide

Research direction

Read datafusion/sqllogictest/src/test_context.rs around the linked function, then inspect datafusion/sqllogictest/bin/sqllogictests.rs around the linked option. Determine how to generate randomized valid configurations and how the run count should be passed and defaulted. Done means sqllogictests can repeat runs with randomized configurations while retaining a default of one run.

Written by the indexing model from the issue text.

Assessment

Tech stack
rust
Domain
testing-qa
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
45/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.