Replace isNullFunctionPointer with real weak-linking support

[Jer Noble]

Replace isNullFunctionPointer with real weak-linking support

[Jer Noble]

Created attachment 399065 [details]
Patch

[Sam Weinig]

What this change adds is what is commonly called "soft" linking, at least in macOS parlance, not weak linking. This doesn't seem like the right direction to go in to me.

[Jer Noble]

(In reply to Sam Weinig from comment #2)
> What this change adds is what is commonly called "soft" linking, at least in
> macOS parlance, not weak linking. This doesn't seem like the right direction
> to go in to me.

There's a crucial difference: this patch doesn't load symbols at runtime; it only checks whether those symbols were loaded at process launch.

[Jer Noble]

Created attachment 399269 [details]
Patch

[Jer Noble]

Created attachment 399273 [details]
Patch

[Sam Weinig]

Comment on attachment 399273 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=399273&action=review

> Source/WebCore/platform/mediastream/libwebrtc/LibWebRTCProviderCocoa.cpp:37
> +WTF_WEAK_LINK_FORCE_IMPORT(webrtc::setApplicationStatus);

Interesting. I couldn't get this to work when I tried.

[EWS]

Committed r261651: <https://trac.webkit.org/changeset/261651>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 399273 [details].

[Radar WebKit Bug Importer]

<rdar://problem/63198533>

(In reply to Sam Weinig from comment #6)
> Comment on attachment 399273 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=399273&action=review
> 
> > Source/WebCore/platform/mediastream/libwebrtc/LibWebRTCProviderCocoa.cpp:37
> > +WTF_WEAK_LINK_FORCE_IMPORT(webrtc::setApplicationStatus);
> 
> Interesting. I couldn't get this to work when I tried.

I think that thing that doesn’t work is redeclaring member functions like this. webrtc ::setApplicationStatus isn’t a member function.