[webkit-reviews] review granted: [Bug 54309] Add a SyncMessageState class to CoreIPC::Connection : [Attachment 82168] Patch

[bugzilla-daemon at webkit.org]

Adam Roben (aroben) <aroben at apple.com> has granted Anders Carlsson
<andersca at apple.com>'s request for review:
Bug 54309: Add a SyncMessageState class to CoreIPC::Connection
https://bugs.webkit.org/show_bug.cgi?id=54309

Attachment 82168: Patch
https://bugs.webkit.org/attachment.cgi?id=82168&action=review

------- Additional Comments from Adam Roben (aroben) <aroben at apple.com>
View in context: https://bugs.webkit.org/attachment.cgi?id=82168&action=review

> Source/WebKit2/Platform/CoreIPC/Connection.cpp:59
> +    RunLoop* m_runLoop;
> +
> +    typedef HashMap<RunLoop*, SyncMessageState*> SyncMessageStateMap;
> +    static SyncMessageStateMap& syncMessageStateMap()
> +    {
> +	   DEFINE_STATIC_LOCAL(SyncMessageStateMap, syncMessageStateMap, ());
> +	   return syncMessageStateMap;
> +    }
> +
> +    static Mutex& syncMessageStateMapMutex()
> +    {
> +	   DEFINE_STATIC_LOCAL(Mutex, syncMessageStateMapMutex, ());
> +	   return syncMessageStateMapMutex;
> +    }

I personally don't like function members to follow data members.

> Source/WebKit2/Platform/CoreIPC/Connection.h:263
> +    class SyncMessageState;
> +    friend class SyncMessageState;

I didn't think the first declaration was necessary. But maybe it is so that the
compiler knows it's a nested class?