Bug 41477 – Direct calls to cast operator in WebFrameImpl.cpp

• Component: Forms
• Status: RESOLVED FIXED

[David Holloway]

Overview: Poor C++ style with direct calls to cast operator in WebFrameImpl.cpp.

Steps to Reproduce:

1) Open WebFrameImpl.cpp

2) Inspect WebKit::WebFrameImpl::registerPasswordListener and WebKit::WebFrameImpl::notifiyPasswordListenerOfAutocomplete

Actual Results: Notice the calls to:

   RefPtr<HTMLInputElement> element = inputElement.operator PassRefPtr<HTMLInputElement>();

   This is using an explicit call to a PassRefPtr cast operator to bridge from a WebInputElement back to an HTMLInputElement.

Expected Results:  Better style is to "unwrap" the WebInputElement explicitly, eg.:

   RefPtr<HTMLInputElement> element(inputElement.unwrap<HTMLInputElement>());

[David Holloway]

Follow up for bug 41236.

[David Holloway]

Created attachment 60272 [details]
Proposed patch.

[David Holloway]

Created attachment 60277 [details]
Proposed patch.

[Darin Fisher (:fishd, Google)]

Comment on attachment 60277 [details]
Proposed patch.

Thanks!  R=me

[WebKit Commit Bot]

Comment on attachment 60277 [details]
Proposed patch.

Clearing flags on attachment: 60277

Committed r62349: <http://trac.webkit.org/changeset/62349>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[WebKit Review Bot]

http://trac.webkit.org/changeset/62349 might have broken Chromium Linux Release

[David Levin]

Rolled out due to breakage on Linux.

http://trac.webkit.org/changeset/62351

[David Levin]

And OSX bustage. Here's the compile error:


/Users/cltbld/Desktop/BuildSlaveData/WebKit-BuildSlave/chromium-mac-release/build/WebKit/chromium/src/WebFrameImpl.cpp: In member function ‘virtual void WebKit::WebFrameImpl::notifiyPasswordListenerOfAutocomplete(const WebKit::WebInputElement&)’:
/Users/cltbld/Desktop/BuildSlaveData/WebKit-BuildSlave/chromium-mac-release/build/WebKit/chromium/src/WebFrameImpl.cpp:1974: error: passing ‘const WebKit::WebInputElement’ as ‘this’ argument of ‘T* WebKit::WebNode::unwrap() [with T = WebCore::HTMLInputElement]’ discards qualifiers

[David Holloway]

Created attachment 60792 [details]
Proposed patch.  Fixes const-ness issues.

Grrrr.  False positive my end.  New patch addresses const-ness issues and pushes the const-ness of the |inputElement| parameter down a level to WebFrameImpl::getPasswordListener().  We need a const_cast at this level due to the use of non-const RefPtr<HTMLInputElement> used as the key to the |m_passwordListeners| map.

[WebKit Commit Bot]

Comment on attachment 60792 [details]
Proposed patch.  Fixes const-ness issues.

Clearing flags on attachment: 60792

Committed r62770: <http://trac.webkit.org/changeset/62770>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.