[Webkit-unassigned] [Bug 83486] [EFL][DRT] EventSender needs implementation of leapForward function

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=83486





--- Comment #10 from Alexander Shalamov <alexander.shalamov at gmail.com>  2012-05-07 12:36:43 PST ---
(In reply to comment #8)
> (From update of attachment 136723 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=136723&action=review
> 
> > LayoutTests/ChangeLog:9
> > +        Reviewed by NOBODY (OOPS!).
> 
> This line usually comes before the change's longer description.

Fixed

> > Tools/ChangeLog:8
> > +        Reviewed by NOBODY (OOPS!).
> 
> Ditto.

Fixed

> > Tools/DumpRenderTree/efl/EventSender.cpp:69
> > +    ulong delay;
> 
> This value is always set from an int below, so why is it a ulong here?

Fixed

> > Tools/DumpRenderTree/efl/EventSender.cpp:75
> > +static const int gDelayedQueueSize = 1024;
> > +static DelayedEvent gEventQueue[gDelayedQueueSize];
> > +static unsigned gStartOfEventQueue;
> > +static unsigned gEndOfEventQueue;
> 
> Why do all this manual bookkeeping instead of using a WTF::Vector?

Fixed (Tried to follow style:)

> > Tools/DumpRenderTree/efl/EventSender.cpp:154
> > +    MouseEventInfo(EvasMouseEvent event, EvasKeyModifier modifiers)
> > +        : event(event)
> > +        , modifiers(modifiers)
> > +        , button(EvasMouseButtonNone)
> > +        , horizontalDelta(0)
> > +        , verticalDelta(0)
> > +    {
> > +    }
...
> 
> This set of constructors could be replaced by a single one:

Fixed


>     MouseEventInfo(EvasMouseEvent event, EvasKeyModifier modifiers, EvasMouseButton button = EvasMouseButtonNone, int horizontalDelta = 0, int verticalDelta = 0)
> 
> > Tools/DumpRenderTree/efl/EventSender.cpp:659
> > +static void feedOrQueueMouseEvent(MouseEventInfo* eventInfo, bool shouldFeedQueuedEvents)
> 
> It's not clear to me when a function is supposed to pass true or false to shouldFeeedQueuedEvents. Can you elaborate?

All mouse events except move should fire all events that are in the queue (same as in other ports).
Usually, test code would programatically press mouse button and then send move command to select or drag element and when button is released, all events with their delays would be re-played in order to simulate user actions.

> > Tools/DumpRenderTree/efl/EventSender.cpp:661
> > +    if ((gDragMode || gStartOfEventQueue != gEndOfEventQueue || gEventQueue[gEndOfEventQueue].delay)
> 
> Ditto about the values of gDragMode (ie. why it is in this check).

Removed, not required at the moment.


> > Tools/DumpRenderTree/efl/EventSender.cpp:688
> > +    // FIXME: We need to pass additional information with our events, so that
> > +    // we could construct correct PlatformWheelEvent. At the moment, max number
> > +    // of clicks is 3
> 
> Are there tests which are broken due to this?

Yes. fast/events/click-count.html
WebKit doesn't care about 1-2-3-ple clicks.
It needs click count, right now, in EFL max number of clicks is 3. GTK has same
limitation, but guys have implemented custom click counter that fixes that issue.

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