[Webkit-unassigned] [Bug 126384] Proxied WebSockets with soup network backend

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=126384





--- Comment #6 from Dirkjan Ochtman <dirkjan at ochtman.nl>  2014-01-03 06:51:36 PST ---
(In reply to comment #5)
> (From update of attachment 220284 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=220284&action=review
> 
> > WebCore/platform/network/soup/SocketStreamHandleSoup.cpp:-94
> > -    if (url.protocolIs("wss"))
> > -        g_socket_client_set_tls(socketClient.get(), TRUE);
> 
> Why are you removing this code? Doesn't this break secure web sockets?

I doesn't seem to in my testing. Before my patch, SocketStreamHandles rely on the pretty low-level GSocketClient API. By using more libsoup infrastructure, this kind of thing is handled for us. In this case, I think (but have not verified) that libsoup can figure out from the URL that TLS support is required.

> 
> > WebCore/platform/network/soup/SocketStreamHandleSoup.cpp:94
> > +    soup_session_proxy_connect(session, uri, connectedCallback, m_id);
> 
> Is this the only change related to proxy support?

Yes, this is the core of the motivating change.

> > WebCore/platform/network/soup/SocketStreamHandleSoup.cpp:238
> > -static void connectedCallback(GSocketClient* client, GAsyncResult* result, void* id)
> > +static void connectedCallback(GIOStream *iostream, GError *error, void* id)
> 
> Is the GIOStream here actually a GSocketClient? If so, perhaps it would make more sense to cast it and avoid the rest of your changes.

Probably! IIRC danw suggested the use of GIOStream in the soup_session_proxy_connect() API. It seems to me like maybe GSocketConnection makes more sense there. If not, then casting on the WebKit side would be fine as far as I'm concerned.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.