[Webkit-unassigned] [Bug 104540] New: [Qt] JS bridge does not transmit QVariants anymore in Qt5

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=104540

           Summary: [Qt] JS bridge does not transmit QVariants anymore in
                    Qt5
           Product: WebKit
           Version: 528+ (Nightly build)
          Platform: PC
        OS/Version: Windows 7
            Status: UNCONFIRMED
          Severity: Blocker
          Priority: P1
         Component: WebKit Qt
        AssignedTo: webkit-unassigned at lists.webkit.org
        ReportedBy: bugzilla at kaffeeschluerfer.com


Windows 7 and Windows XP
Visual Studio 2010
Qt5 RC
(I can not test if any other platforms are affected.)

In Qt 4.7.x and 4.8.x JSON object were transmitted as QVariant into the C++ Slot:


---JavaScript extract:---
var jsonTest = {
    service: "LogService",
    type: "request",
    methodName: "logIt",
    count: performCounter
};
this.Bridge.slotQVariant(jsonTest);


---C++ extract---
void JavaScriptConnection::slotQVariant(QVariant data)
{
   //... data can be converted to QVariantMap
}

This worked fine, but with Qt5 we get the error 'Trying to construct an instance of an invalid type, type id: 732670891'. It helps to make in JS a 'JSON.stringify(jsonTest)' and to expect a QString in the C++ Slot:

---JavaScript extract:---
var jsonTest = {
    //... see above
};
this.Bridge.slotQString(JSON.stringify(jsonTest));


---C++ extract---
void JavaScriptConnection::slotQString(QString data)
{
   //...
}


However, this behavior change seems not correct, because (first) it breaks compatability for no reason and (second) it has performance drawbacks on C++ side, because the QString has to be converted.
[Furthermore there's a severe speed decrease, and I do not mean the mentioned conversion, when using the bridge several 10000s of times, but I will post this as a different bug]

As attachment I added a VS2010 project. It's stripped down to the bare problem, so you can hopefully and quickly see the problem. Only the Release build is configured correctly. There is some time measurement on C++ side, but this is of no relevance here.

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