I was working on optimising Subversion. One of its bottleneck is computing file hashes. First thing I tried was to use OpenSSL for that. It significantly improved its performance. It made it 8x faster in some cases.
However, it was not optimal to use it:
- OpenSSL is heavy and bloated. It's like a whole operating system on its own.
- Since OpenSSL 3.0, low-level digest APIs are deprecated, and instead they recommend to use EVP which might be slow in some cases.
So I decided to create a new library for computing cryptographic digests. It basically takes OpenSSL's implementations, but with an aim to provide simple and lightweight access to popular hash algorithms with a focus on performance.
However, it was not optimal to use it:
- OpenSSL is heavy and bloated. It's like a whole operating system on its own.
- Since OpenSSL 3.0, low-level digest APIs are deprecated, and instead they recommend to use EVP which might be slow in some cases.
So I decided to create a new library for computing cryptographic digests. It basically takes OpenSSL's implementations, but with an aim to provide simple and lightweight access to popular hash algorithms with a focus on performance.