Jonathan Lennox

I feel like there should be some infra-side configuration about whether we want to do ssrc-rewriting, not just the client capability (unless sending the client capability is triggered by some...

This seems good to me, but should wait to land until the JVB part is done. (We'll need to make sure JVB and Jicofo are updated in sync with each...

It's getting a failure in the unit tests, do we understand why?

Some of these are only "test"-scope dependencies, and as such shouldn't introduce any issues of version conflicts, project size, or license issues for consumers of the library. (`mvn dependency:tree -Dscope=compile`...

> [openssl/openssl#17811](https://github.com/openssl/openssl/issues/17811) --- looks like need to either store the key or else only support 3.0.3 and greater :( . @bifurcation thoughts ? Storing the key and re-applying it will...

Couldn't the ifdef that sets `SRTP_OSSL_USE_EVP_MAC` be conditional on being OpenSSL 3.0.3 or later, and just use the HMAC-flavored functions for OpenSSL

The other possibility - which I think should work on 3.0.2 without a performance penalty, though I haven't tried it - would be to have srtp_hmac_alloc store a "pristine" MAC...

> @JonathanLennox I added the dup function but the performance hit is noticeable. It will only use this based on a runtime check of the version otherwise it will reinit...

Note that even though these functions are deprecated, they're still fully available in OpenSSL 3.x, and the high-level APIs just call these low-level functions directly. So this change wouldn't (as...

I agree that srtp_crypto_kernel_do_load_cipher_type will do the right thing; the problem is that a bunch of the code in srtp.c to set up the GCM modes is conditionally compiled under...