Enough of this libwebrtc.dylib nonsense. It isn't worth having a complicated production build system that only one person understands.
Created attachment 305346 [details] patch
Comment on attachment 305346 [details] patch View in context: https://bugs.webkit.org/attachment.cgi?id=305346&action=review > Source/WebCore/ChangeLog:12 > + * Configurations/WebCoreTestSupport.xcconfig: Should this eliminate the code in Source/WebCore/platform/mediastream/libwebrtc/LibWebRTCProvider.cpp that attempts to dlopen("libwebrtc.dylib")?
I believe that there are Apple-internal reasons for libwebrtc being a dylib, seen in <rdar://problem/30470169&30470847>.
I would do that if we cannot upstream the WEBRTC_DYLIB_EXPORT changes to libwebrtc. Given where we are, I do not have a strong opinion, although that would simplify the build system which is a good thing. Also, if we are doing so, the webcore binary size will increase a bit. Some code would be removed like webrtAvailable check.
We need to discuss the Apple-internal implications (not in these bug comments, but Apple-internally) before we make the call on this.