Because this seems to be extremely slow compared, to say, OpenSSL. While clearly it would benefit from optimizations similar to what fastpbkdf2 uses, biggest bottleneck seems to be GNAT.SHA* modules. Ideally, I would prefer to have a separate crate for sha and hmac that would implement these algorithms in a platform-independent, portable and hopefully faster manner.
Because this seems to be extremely slow compared, to say, OpenSSL. While clearly it would benefit from optimizations similar to what fastpbkdf2 uses, biggest bottleneck seems to be GNAT.SHA* modules. Ideally, I would prefer to have a separate crate for sha and hmac that would implement these algorithms in a platform-independent, portable and hopefully faster manner.