Ask HN: Why not to build an open-source C++ project on Intels oneTBB?

Intel's thread building blocks and OpenMP seems to be dominant library for accelerating single threaded code which has some parallelizable sections. The nested parallelism in the open source project has makes me discount OpenMP based solutions. None of the core maintainers has experience with raw parallel C++ programming using user created and managed threads and locking or lock-free data structures. The code has no BLAS style operations. We target workstation or server class devices, embedded support is not planned.

Under those circumstances TBB looks interesting and i am favoring that solution. Is there a reason why an open source project should not use Intels TBB?

What alternatives should be considered?