| Issue |
170178
|
| Summary |
Remove or enforce test uniqueness requirement in the LLDB API test suite
|
| Labels |
lldb,
test-suite
|
| Assignees |
|
| Reporter |
JDevlieghere
|
Currently, the LLDB API test suite requires that test names (i.e. the name of the Python test class) are unique. However, this requirement only gets enforced when using `lldb-dotest` as the test runner (i.e. not `llvm-lit`). Some LLDB developers prefer using `lldb-dotest` during development for a variety of (good) reasons. However, `llvm-lit` is the canonical test runner that's used in CI and when using the `check-lldb*` targets, which is why we don't discover these name collisions sooner.
Naming collisions in tests pop up periodically and prompts the question why this keeps happening. There are two ways we can improve the current situation:
1. Move the check that catches this and bail out early. This way we uncover naming collisions as soon as they're introduced.
2. Drop the requirement. Based on Pavel's comment in https://github.com/llvm/llvm-project/pull/137681 it seems like the reason for its existence is no longer relevant.
We didn't reach consensus in https://github.com/llvm/llvm-project/pull/137681 so let's use this issue to settle on a direction.
![]()
_______________________________________________
llvm-bugs mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
