It makes non-masquerading objects boolify to False instead of True.
Created attachment 254927 [details] the patch
Comment on attachment 254927 [details] the patch View in context: https://bugs.webkit.org/attachment.cgi?id=254927&action=review > Source/JavaScriptCore/ChangeLog:8 > + * ftl/FTLLowerDFGToLLVM.cpp: Put this in an anonymous namespace. We should have done that all along. It makes it easier to add debug code. Should we be using anonymous namespaces throughout WebKit instead of “static” to make things local to a file? A while back this made debugging harder, which I think is why we didn’t do that at the time.
<rdar://problem/21393673>
(In reply to comment #2) > Comment on attachment 254927 [details] > the patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=254927&action=review > > > Source/JavaScriptCore/ChangeLog:8 > > + * ftl/FTLLowerDFGToLLVM.cpp: Put this in an anonymous namespace. We should have done that all along. It makes it easier to add debug code. > > Should we be using anonymous namespaces throughout WebKit instead of > “static” to make things local to a file? A while back this made debugging > harder, which I think is why we didn’t do that at the time. I didn't know about the debugging issue. Compiler debugging usually involves glorified printf()'s. My preference is to use anonymous namespaces instead of static, because it allows us to create one block in a module that contains all of the module-local stuff - functions, classes, variables, enums, etc. It's very liberating - you can use short names. We've been steadily converting the DFG/FTL to use anonymous namespaces for the module-local classes. Each compiler phase usually involves at least one module-local class. That's really all I'm doing here.
Landed in http://trac.webkit.org/changeset/185600