It looks like the failures are from the changes to check for --offload-arch and -fsycl-targets together without --offload-new-driver. The usage triggering is the alias for --cuda-gpu-arch. Maybe forgo the error diagnostic and emit a warning about 'option unused' (i.e. only claim the --offload-arch usage when we know the value is being used)

Oops - didn't mean to close :)

This pull request is stale because it has been open 180 days with no activity. Remove stale label or comment or this will be automatically closed in 30 days.

This pull request was closed because it has been stalled for 30 days with no activity.