Fix clientX/Y properties of touch pointer events

third_party/WebKit/LayoutTests/virtual/pointerevent/fast/events/pointerevents/pointer-event-properties-in-iframe-expected.txt

Index: third_party/WebKit/LayoutTests/virtual/pointerevent/fast/events/pointerevents/pointer-event-properties-in-iframe-expected.txt
diff --git a/third_party/WebKit/LayoutTests/virtual/pointerevent/fast/events/pointerevents/pointer-event-properties-in-iframe-expected.txt b/third_party/WebKit/LayoutTests/virtual/pointerevent/fast/events/pointerevents/pointer-event-properties-in-iframe-expected.txt
new file mode 100644
index 0000000000000000000000000000000000000000..09a12bcae0895bd832d00bccc9c2c416370d959b
--- /dev/null
+++ b/third_party/WebKit/LayoutTests/virtual/pointerevent/fast/events/pointerevents/pointer-event-properties-in-iframe-expected.txt
@@ -0,0 +1,37 @@
+
+This test verifies clientX/Y of pointer events inside iframe.
+
+On success, you will see a series of "PASS" messages, followed by "TEST COMPLETE".
+
+
+PASS successfullyParsed is true
+
+TEST COMPLETE
+pointerdown of touch is recieved:
+clientX = 150
+clientY = 100
+pointermove of touch is recieved:
+clientX = 150
+clientY = 100
+pointerup of touch is recieved:
+clientX = 150
+clientY = 100
+pointerdown of touch is recieved:
+clientX = 150
+clientY = 100
+pointermove of touch is recieved:
+clientX = 150
+clientY = 100
+pointerup of touch is recieved:
+clientX = 150
+clientY = 100
+pointerdown of touch is recieved:
+clientX = 10
+clientY = -107

[bokan, 2016/05/09 21:03:07]

This looks wrong to me. For starters, at 2X zoom the iframe is at (100px, 200px)
in DIPs so I think the touch is occurring at (50px, 0px) in CSS pixels in the
iframe's client coordinates. We should certainly not be seeing negative client
coordinates here.

Touching/clicking at the bottom right corner of the 300x300 iframe should give
client coordinates of (299,299) CSS pixels regardless of zoom level.

That said, I'm assuming eventSender coordinates are "renderer coordinates", so
they'd be in DIPs relative to the window origin. That makes most sense to me but
it's not an assumption I've confirmed. This would mean that as you zoom in,
you'd have to increase the coordinates given to eventSender to hit the bottom
right of the iframe.

FYI, DIP / pageZoom = CSS pixel for this case.

[Navid Zolghadr, 2016/05/10 15:01:40]

I was looking at the implementation of contentsToFrame function. It seems that
it doesn't consider the zoom factor. I couldn't find another function that also
considers that.
Is that okay that I do it manually once again (basically the same way it was in
the first patch)?

[bokan, 2016/05/10 16:33:53]

Ah, touchInfo is already zoomed...we should only be applying zoom factor just
before we provide it to the web contents. Internally, everything should be in
unzoomed DIPs. lgtm to do this manually since this is all in flux (you mentioned
getting rid of TouchInfo, right) but we should avoid making manual conversions
like this long term.

[Navid Zolghadr, 2016/05/10 17:46:42]

Done.

+pointermove of touch is recieved:
+clientX = 10
+clientY = -107
+pointerup of touch is recieved:
+clientX = 10
+clientY = -107
+