Daniil Goncharov

Hi @aantron, Both libraries solve similar problems, but with a different implementation. Difference: Better Enums * Need registered enum (by macro). * C++98/11. Magic Enums * No need registered enum....

@dpacbach Unfortunately, "magic" depends on the compiler :D

C++20 coroutines?

In c++11 maybe we can use `std::async`

`for_each` was made by a method, not a free function. I'm not sure what's best. It is possible to make functions `begin()`/`end()`, but to do this will have to create...

Thank you for your feedback. Yes, I will try to make a solution with `begin()`/`end()`. I'll update this PR when I have something to show.

@zengzhiw Hi, my build https://github.com/Neargye/hello_tf_win_c_api, maybe this will help.

@WangDeyu Hi, my example https://github.com/Neargye/hello_tf_win_c_api, maybe this will help.

Try this ``` ... TF_Operation* train_ = TF_GraphOperationByName(graph_, "train_step"); std::vector target_opers = {train_ }; ... TF_SessionRun(sess_, nullptr, // Run options. &input_op_, &input_tensor_, 1, // Input tensors, input tensor values, number...

TF_SessionRun takes an array of target, so the type const TF_Operation* const*.