Low risk non-nullable => nullable change

third_party/WebKit/Source/core/html/HTMLInputElement.idl

Index: third_party/WebKit/Source/core/html/HTMLInputElement.idl
diff --git a/third_party/WebKit/Source/core/html/HTMLInputElement.idl b/third_party/WebKit/Source/core/html/HTMLInputElement.idl
index 669c8e19aa5d3d984cf719d87fa4aa9e324c7342..c2c5587f44013ea0e13f59c4140e3afef2c20b31 100644
--- a/third_party/WebKit/Source/core/html/HTMLInputElement.idl
+++ b/third_party/WebKit/Source/core/html/HTMLInputElement.idl
@@ -80,9 +80,9 @@ interface HTMLInputElement : HTMLElement {
 
     void select();
     // FIXME: selectionStart and selectionEnd should be unsigned long.
-    [RaisesException, ImplementedAs=selectionStartForBinding] attribute long selectionStart;
-    [RaisesException, ImplementedAs=selectionEndForBinding] attribute long selectionEnd;
-    [RaisesException, ImplementedAs=selectionDirectionForBinding] attribute DOMString selectionDirection;
+    [RaisesException, ImplementedAs=selectionStartForBinding] attribute long? selectionStart;

[foolip, 2016/12/13 22:37:31]

OK, so I said that making readonly attributes nullable doesn't change the
generated code, but it's not always true. For things like WTF::String and
pointers that can be null, it just works, but for integers etc. an extra
argument is added to the getter, which is why the compile is now failing.

I'd exclude this change, because it would also need to actually return null in
some situation for it to be useful, which then requires tests and so on.

[lunalu1, 2016/12/14 11:00:14]

Done.

+    [RaisesException, ImplementedAs=selectionEndForBinding] attribute long? selectionEnd;
+    [RaisesException, ImplementedAs=selectionDirectionForBinding] attribute DOMString? selectionDirection;
     [RaisesException] void setRangeText(DOMString replacement);
     [RaisesException] void setRangeText(DOMString replacement,
                                         unsigned long start,