Chromium port requires Frame* object to share information in the frame. This change is a port dependent part of the change for 78459.
Created attachment 126722 [details] Patch
Comment on attachment 126722 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=126722&action=review > Source/WebCore/ChangeLog:8 > + Add owner Frame* as an argument to SocketStreamHandle::create, etc > + https://bugs.webkit.org/show_bug.cgi?id=78460 > + > + Reviewed by NOBODY (OOPS!). > + > + No new tests because of just adding an unsed argument. Please write reasons why you add an unused argument. > Source/WebCore/websockets/WebSocketChannel.cpp:143 > + Document* document = static_cast<Document*>(m_context); Is this cast safe? Can't m_context be a worker?
Created attachment 126730 [details] to show overview to dependent change reviewers
Sorry, please just ignore previous attachment. I'll post it to another bug entry, but mistakenly post here.
Comment on attachment 126722 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=126722&action=review >> Source/WebCore/ChangeLog:8 >> + No new tests because of just adding an unsed argument. > > Please write reasons why you add an unused argument. OK, I added. >> Source/WebCore/websockets/WebSocketChannel.cpp:143 >> + Document* document = static_cast<Document*>(m_context); > > Is this cast safe? Can't m_context be a worker? Yes, currently m_context always has a Document. But, it could cause a security risk for future changes. I'll find a better way.
> Add owner Frame* as an argument to SocketStreamHandle::create, etc Please don't. This would be an extreme layering violation.
Ah... I see. I should not make a dependency on WebCore in platform layer, right? I'll find another way.
(In reply to comment #7) > I should not make a dependency on WebCore in platform layer, right? Right. The code in platform/ must not depend on files in other directories.