Rich Salz
Have you proofed this design out against things like BBR and L4S?
I asked a question two months ago that has not been answered. https://github.com/openssl/openssl/pull/18018#issuecomment-1105193404
>the concept of changeables should allow for these modern CC methods Okay, so what CC algorithms are supported by the basic API? Reno? Cubic? "should" is weak. If the project...
> The API is not set in stone. And yet #18676 would add API's (and code) that basically cement this design; that PR claims this has already been "workshopped" (whatever...
Sure, it's just code. But the design doesn't even understand the main congestion systems used in the short term, and you don't seem to be participating in the design of...
cmake would not address the "perl is needed" issue. look at all the .pl scripts, such as those that generate *all* the assembler code and those in util directory. Even...
> Compilers are getting better and better. Some of the assembly might be slower than the hand rolled assembly. Nope. And there are security issues -- AES constant time (hopefully...
I think the true answer is that the project does not know every platform that uses OpenSSL. There are many embedded devices, for example, and breaking them would be bad...
@newpavlov you are right, it does! I wonder if that should be added to the crypto list?
yeah for zulip! but it needs a host.