ResourceRequest is forward declared as class and the actual definition is class. This will cause WINSCW linker to fail all function calls where ResourceRequest is one of the parameters
Created attachment 41653 [details] proposed fix for ResourceLoaderNotifier.h
Created attachment 41654 [details] proposed fix for ResourceLoaderNotifier.h take 2 Added the wrong patch. that will be coming up shortly :)
Comment on attachment 41654 [details] proposed fix for ResourceLoaderNotifier.h take 2 > #include <wtf/Noncopyable.h> > - ^^^ we don't add whitespace like this... > namespace WebCore { > > class AuthenticationChallenge; > @@ -40,7 +39,7 @@ class Frame; > class ResourceError; > class ResourceLoader; > class ResourceResponse; > -class ResourceRequest; > +struct ResourceRequest; > class ScriptString; Yes and No. CF, Chromium, CURL, SOUP refer to it as struct ResourceRequest, mac is having a proper class. WebCore/platform/network/mac/ResourceRequest.h:41: class ResourceRequest : public ResourceRequestBase { I think it is time to change the other ports to class ResourceRequest as well because the base class is declared as a "class".
(In reply to comment #3) > WebCore/platform/network/mac/ResourceRequest.h:41: class ResourceRequest : > public ResourceRequestBase { > > I think it is time to change the other ports to class ResourceRequest as well > because the base class is declared as a "class". Thanks for the review, I'll prepare a patch that changes all forward declarations and the struct definition to class. This time without additional line removal :) Patch will be slightly bigger as we touch a lot of files (75 at quick glance).
Created attachment 42084 [details] unification of ResourceRequest class/struct usage
Do you mind if I change the changelog subject line to "Change ResourceRequest to be ..."? Patch looks fine, I will try to build Gtk, Qt and Mac and then land it.
(In reply to comment #6) > Do you mind if I change the changelog subject line to "Change ResourceRequest > to be ..."? Patch looks fine, I will try to build Gtk, Qt and Mac and then land > it. No, I don't mind. Go ahead and change it.
Comment on attachment 42084 [details] unification of ResourceRequest class/struct usage Should be compiled tested before landing...
bug 28003 is yet another "dupe" of this.
Comment on attachment 42084 [details] unification of ResourceRequest class/struct usage the commit-queue will test the compile before it lands.
Comment on attachment 42084 [details] unification of ResourceRequest class/struct usage Clearing flags on attachment: 42084 Committed r50625: <http://trac.webkit.org/changeset/50625>
All reviewed patches have been landed. Closing bug.