Hi Christian,
I *think* this interface might be the wrong one, as it makes it impossible to pass options to the test command.
You might also need to handle environment variables. I suggest testing your implementation out with rccl. Although, I suppose one could always use a wrapper.
One thing that might need some consideration eventually is that not all software is compatible with all GPU architectures. The hipBLASLt and rocWMMA libraries would be good examples of this. We may eventually need a way to express that sort of constraint.
I'm leaning towards (B) because when I think of the autopkgtests I implemented, it was more typical for one test call to accept options, than to call more than one test (runner). Hence, if a wrapper cannot be avoided, then best to relegate it to the rarer use case.
That seems reasonable to me. There might be fancier options, but
this would be sufficient to ensure that the wrapper script is a
simple program. And we may be able to provide alternative modes of
operation in the future to handle other common cases, if need be.
I believe I may be able to avoid needs-sudo. [....] The nice thing is that both of this would also work for podman workers, where we currently don't save this important information.
That is great!
Sincerely,
Cory Bloor