Do not use type aliases in CTAD declarations

[tomilov]

Result type of class template argument deduction declaration must be explicit class template specialization, not one mediated by alias template or type alias, like it is here.

The code has mistake, that leads to hard error during compilation:

/usr/include/tbb/concurrent_hash_map.h(1247): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_hash_map.h(1253): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_priority_queue.h(536): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_priority_queue.h(540): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_unordered_map.h(282): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_unordered_map.h(287): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_unordered_map.h(292): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_unordered_map.h(297): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_unordered_map.h(464): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_unordered_map.h(469): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_unordered_map.h(474): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_unordered_map.h(479): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_unordered_set.h(235): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_unordered_set.h(240): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_unordered_set.h(245): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_unordered_set.h(250): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_unordered_set.h(420): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_unordered_set.h(425): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_unordered_set.h(430): error: the return type must directly designate a specialization of the associated class template
/usr/include/tbb/concurrent_unordered_set.h(435): error: the return type must directly designate a specialization of the associated class template

Compilers are:

• clang version 10.0.1
• g++ (GCC) 10.2.0

Both failed when running from within /opt/cuda/bin/nvcc -std=c++17:

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2020 NVIDIA Corporation
Built on Wed_Jul_22_19:09:09_PDT_2020
Cuda compilation tools, release 11.0, V11.0.221
Build cuda_11.0_bu.TC445_37.28845127_0

[kboyarinov]

Hi, @tomilov ,

Deduction guides for oneTBB components were reworked to avoid compilation issues in oneTBB 2021.2.0.
If you still have any issues, feel free to contact us.

[tomilov]

Sadly I cannot check now, but I trust you. Thank you.