test(core): add tests for versioning

[meteorgan]

Which issue does this PR close?

part of: #2611

Rationale for this change

What changes are included in this PR?

Are there any user-facing changes?

[meteorgan]

In #5106, I didn't address the DeleteMarker in the response. However, it could be useful for the users who want to know which objects haven been deleted. Should we return the DeleteMarker as a object version with a delete_marker tag when performing a list_with_version request ?

[Xuanwo]

In #5106, I didn't address the DeleteMarker in the response. However, it could be useful for the users who want to know which objects haven been deleted. Should we return the DeleteMarker as a object version with a delete_marker tag when performing a list_with_version request ?

I believe they should be included when users have list_with(path).deleted(true). Would you like to start another issue to track this?

[meteorgan]

In #5106, I didn't address the DeleteMarker in the response. However, it could be useful for the users who want to know which objects haven been deleted. Should we return the DeleteMarker as a object version with a delete_marker tag when performing a list_with_version request ?

I believe they should be included when users have list_with(path).deleted(true). Would you like to start another issue to track this?

Ok. And I want to discuss some additional details about this issue.

[meteorgan]

If versioning is not enabled, should we return ErrorKind::Unsupported for every operation with version parameter ? These cases weren't addressed in earlier PR, such as: #4873

[Xuanwo]

If versioning is not enabled, should we return ErrorKind::Unsupported for every operation with version parameter ? These cases weren't addressed in earlier PR, such as: #4873

I believe we should make our best effort to fetch them since it doesn't affect the result. It's fine to attempt fetching even if versioning isn't enabled. The situation is different with write_with_if_none_match, which may lead to incorrect behavior like writing data which wrong condition.

[meteorgan]

If versioning is not enabled, should we return ErrorKind::Unsupported for every operation with version parameter ? These cases weren't addressed in earlier PR, such as: #4873

I believe we should make our best effort to fetch them since it doesn't affect the result. It's fine to attempt fetching even if versioning isn't enabled. The situation is different with write_with_if_none_match, which may lead to incorrect behavior like writing data which wrong condition.

what about list ? in #5106 (comment), we decided to return error when versioning is not enabled.

[Xuanwo]

what about list ? in #5106 (comment), we decided to return error when versioning is not enabled.

Well, that makes sense. Let's return unsupported for all those cases.

[Xuanwo]

Perfect! Thanks a lot.