[CALCITE-4595] Split DiffRepository test data into per-test files

• Type: Improvement
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 1.26.0
• Component/s: core
• Assignee: Unassigned
• Reporter: Vladimir Sitnikov
• Created: 04/May/21 18:52
• Updated: 11/Feb/23 16:00

Description

The current issues:

• the file is 13'000 lines, so it is hard to browse in IDE
• contributions might result in merge conflicts, and "sorting the items" requires non-standard procedure (e.g. Calcite-specific sorting of the calcite-specific XML files)

If we split the file into several smaller ones, then:

• The number of merge conflicts would be greatly smaller
• We won't need to sort the items inside the files.
• At best, it would be compatible with https://approvaltests.com/ which has IDE plugins to automatically update the reference files
• plan changes in GitHub PR would convey the test name right in the filename, so the review would be easier

The drawbacks of having more files might be:
a) Each file might have to have a license header that consumes ~16 lines. We could either blindly accept it or exclude it from RAT check.
b) Global plan changes would touch lots of XML files. Currently, we copy a couple of files from "actual to expected", and with the new approach, we might need to copy more files. However, copying several files is not that hard.

Issue Links

• is related to: CALCITE-4593 DiffRepository tests should fail if XML resources are not in alphabetical order
• is related to: CALCITE-5501 SqlToRelConverterTest.checkActualAndReferenceFiles fails intermittently in Jenkins CI