Consolidate how ImageDocument sets image styling

third_party/WebKit/Source/core/html/ImageDocument.cpp

Index: third_party/WebKit/Source/core/html/ImageDocument.cpp
diff --git a/third_party/WebKit/Source/core/html/ImageDocument.cpp b/third_party/WebKit/Source/core/html/ImageDocument.cpp
index 88826f4d3894c9781ee2ab8eba50a45a760c2b64..569334cb5bec0c2984d5953954523518d3d25b25 100644
--- a/third_party/WebKit/Source/core/html/ImageDocument.cpp
+++ b/third_party/WebKit/Source/core/html/ImageDocument.cpp
@@ -398,6 +398,11 @@ void ImageDocument::updateImageStyle() {
         }
 
         newCheckerSize = round(std::max(1.0, newCheckerSize / scale));
+      } else if (!imageFitsInWindow()) {
+        // In desktop mode, the user can click on the image to zoom in or out.
+        DCHECK_EQ(m_shrinkToFitMode, Desktop);
+        imageStyle.append(m_shouldShrinkImage ? "cursor: zoom-in; "

[pdr., 2016/11/21 23:26:47]

I see we're using "m_imageElement->setInlineStyleProperty(CSSPropertyCursor,
CSSValueZoomOut);" to do this elsewhere. Is the style code in this function
overwriting that? I wonder if we should just have this function be in charge of
managing the image style (i.e., have the other callers of
m_imageElement->setInlineStyleProperty call updateImageStyle())?

[gone, 2016/11/21 23:43:26]

Yeah, on desktop this is called during ::imageLoaded(), after the cursor is
initially set via setInlineStyleProperty(), effectively erasing it.  This
function isn't called anywhere else after initialization, though, and I'm not
sure if it's cheaper overall to edit the whole style tag than to just
remove/change the single style property explicitly when we know it needs to be
updated.  It might make sense to just pull out the cursor-setting code
specifically, then have everything call that.  WDYT?

+                                              : "cursor: zoom-out; ");
       }
 
       // The only thing that can differ between updates is the checker size.