Bug 38304 – [Qt] Reinstate qt_drt_ symbol exports as of QtWebKit 4.6 release

• Status: RESOLVED
• Resolution: FIXED
• Component: WebKit Qt

[Laszlo Gombos]

It would be a small effort (and small overhead) and potentially a big win to maintain some of the previously exported qt_drt_ symbols after r57433 for some time until all QtWebKit clients transition to the API changes. 

There is no backward compatibility grantee on the qt_drt interfaces, but some of these interfaces are only not exported because a "clean" Qt-style interface has not been commonly agreed (and not because they are not useful). 

This is a stop gap solution until a proper QtWebKit API is commonly agreed for some of this functionality.

[Laszlo Gombos]

Created attachment 54660 [details]
proposed patch

[Kenneth Rohde Christiansen]

Comment on attachment 54660 [details]
proposed patch

> void QWEBKIT_EXPORT qt_drt_run(bool b)

I'm pretty sure this one is not used by anyone outside the DRT, does it really make sense to reintroduce it?


WebKit/qt/WebCoreSupport/DumpRenderTreeSupportQt.cpp:431
 +  bool QWEBKIT_EXPORT qt_drt_pauseAnimation(QWebFrame *qframe, const QString &animationName, double time, const QString &elementId)
Wrong coding style, please fix before landing. Also, just use frame, not need to call it qFrame/qframe.

I think it makes sense to order the DRT methods together and the QWebPage ones as well; right now they are mixed.

[Simon Hausmann]

I'm fine with reinstating the qt_* functions that were accidentially removed, but why should we reinstate the qt_drt_* ones? As the DRT suffix suggests they are only meant for the DRT implementation, which doesn't need them.

Are you using the qt_drt_* functions in applications?

[Kenneth Rohde Christiansen]

Judging from the email to the list, they indeed are.

[Laszlo Gombos]

Bug 31875 would be a good example of a functionality that should be a public API but the only way this is exposed today is trough DRT methods.

[Laszlo Gombos]

Landed as http://trac.webkit.org/changeset/58527 with the changes suggested by Kenneth.

I kept qt_drt_run for now - we would need to mark the symbol ABSENT in the def file for Symbian if we remove it. 

It also potentially allows using the same DRT binary for a few different WebKit versions to check for differences between releases, but I'm not sure if this is a compelling enough case.