?hbmv and ?hpmv and the documentation notes seem to be inconsistent

[lzzwsf]

I now need to perform a C row-major-order operation on a vector y with a step incy of 2. According to the formula in the documentation I should get a result of alphaAX+beta*y. but the actual return value does not match the result of manually performing the operation according to this formula. Tracing the memory revealed that CBLAS exchanged the real and imaginary parts of Y before the operation. Meanwhile?hpmv also has this problem, is this by design or a bug?If it's by design, a timely documentation update is needed!

Translated with DeepL.com (free version)