Attention is currently required from: laforge, pespin.

fixeria has posted comments on this change. ( https://gerrit.osmocom.org/c/libosmo-abis/+/35502?usp=email )

Change subject: trau/osmo_ortp: support building without OrtpMemoryFunctions API ......................................................................

Patch Set 1:

(1 comment)

Patchset:

PS1:

• Yes, it's a public dependency and it's part of the public API of the library

How does this contradict what I am saying? I am not saying that `bcmtoolbox` is not a public dependency of `ortp`. I am saying that for libosmo-abis, `bcmtoolbox` is **not a direct** (but indirect or implicit) dependency, pulled by `ortp`. We don't require `bcmtoolbox` in our `configure.ac` nor in `debian/control`. This is why I don't think it's a good idea to use API of `bcmtoolbox`.

...

they may switch to some other library any day.

That would break its ABI/API.

This whole situation with libosmo-abis not building against recent `ortp` indicates that they don't care that much about breaking API/ABI. They also did modify signatures of existing public functions, and this is why we have quirks like `HAVE_ORTP_LOG_DOMAIN` and `RTP_SIGNAL_PTR_CAST`.

I am inclined to remove this talloc-to-ortp integration completely (even for old ortp versions), as was suggested by Hoernchen in the IRC. It's questionable whether we really want to use talloc here.