Bug 79237 – [EFL][DRT] Implement setValueForUser and setAutofilled

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

[Grzegorz Czajkowski]

Add missing implementation setValueForUser and setAutofilled to EFL's LayoutTestController.
Those implementations are related with input fields and allow to pass following tests:
 - fast/forms/onchange-setvalueforuser.html
 - fast/forms/input-autofilled.html
 - fast/forms/reset-autofilled.html

[Grzegorz Czajkowski]

Created attachment 128208 [details]
proposed patch

[Grzegorz Czajkowski]

CC'ing developers involved in LayoutTests/DRT

[Hajime Morrita]

Comment on attachment 128208 [details]
proposed patch

Accessing JSValueRef smells wrong. When you notice you're doing it, it will indicate something wrong.

IMHO, what you need is WebKit level API to access WebCore::Element and its autofill methods.
For example, Mac port has DOMElement obj-c class and Chromium has WebElement as a wrapper of WebCore::Element.
I think you need similar for EFL. Just making bot green doesn't help in this case.

Please consult some expert in your team and discuss what is the good API for EFL port.

[Hajime Morrita]

Point here is that you cannot implement auto-complete feature for your browser with EFL API.
You'll need such an API and DRT should demonstrate it.

[Martin Robinson]

(In reply to comment #4)
> Point here is that you cannot implement auto-complete feature for your browser with EFL API.
> You'll need such an API and DRT should demonstrate it.

DumpRenderTreeSupport is actually a private part of the EFL API that predates WebCore.internals and fulfills much the same purpose. GTK+, Qt, and EFL use it to add private API for the DRT.

[Hajime Morrita]

Comment on attachment 128208 [details]
proposed patch

(In reply to comment #5)
> (In reply to comment #4)
> > Point here is that you cannot implement auto-complete feature for your browser with EFL API.
> > You'll need such an API and DRT should demonstrate it.
> 
> DumpRenderTreeSupport is actually a private part of the EFL API that predates WebCore.internals and fulfills much the same purpose. GTK+, Qt, and EFL use it to add private API for the DRT.

Hmm interesting. In general I understand DumpRenderTree serves such purposes.
But for this case, is it making sense to make the test pass even if the port doesn't support the feature?

Clearing r- since I realized that I don't have enough understanding of the background context.

[Grzegorz Czajkowski]

Comment on attachment 128208 [details]
proposed patch

I added request for the review again.
Martin could you review it because you've already reviewed the similar patch for WebKit-GTK port?

[Andreas Kling]

Comment on attachment 128208 [details]
proposed patch

rs=me as this looks like pretty much exactly like the Gtk+ implementation of the same functionality.

[Grzegorz Czajkowski]

Created attachment 128724 [details]
patch for landing

Rebased to head.

[WebKit Review Bot]

Comment on attachment 128724 [details]
patch for landing

Clearing flags on attachment: 128724

Committed r108806: <http://trac.webkit.org/changeset/108806>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.