dsseng wrote:
> I do think it would be good to have a test case for this; this is tricky code
> (and probably also performance-sensitive), and walking through a test case
> helps to think through the resulting behaviour.
>
> You can have a look at [this test
> case](https://searchfox.org/llvm
https://github.com/dsseng created
https://github.com/llvm/llvm-project/pull/148019
A compile_commands.json with Windows paths would make clangd assert on a
POSIX system, and vice versa.
Make path traversal functions skip paths of the incompatible format.
Could not invent a test for this one, s
