Merge lp://staging/~justinmcp/oxide/cursor-support into lp://staging/~oxide-developers/oxide/oxide.trunk

Thanks for working on this. This is pretty much ok. The only question I have is:

+void RenderWidgetHostView::UpdateCursor(const WebCursor& cursor) {
+ if (GetFormFactorHint() != FORM_FACTOR_DESKTOP) {
+ return; // Cursor only on desktop
+ }
+

Is this check really necessary? Is this done in other parts of the stack that process cursor changes? And do we know whether Chromium calls UpdateCursor() without mouse events?

+ QImage::Format format = QImage::Format_Invalid;
+ switch (cursor_info.custom_image.config()) {
+ case SkBitmap::kRGB_565_Config: format = QImage::Format_RGB16;
+ case SkBitmap::kARGB_4444_Config: format = QImage::Format_ARGB4444_Premultiplied;
+ case SkBitmap::kARGB_8888_Config: format = QImage::Format_ARGB32_Premultiplied;
+ default: ;
+ }

I was going to say that this bit is missing an indent for each of the case statements, but I've just noticed that this rule is already applied quite inconsistently in this file, so it doesn't matter for now :)

Note, I've just updated to Chromium 35.0.1908.4 which changes some things related to this merge :(

- blink::WebCursor is now content::WebCursor
- webkit/common/cursors/webcursor.h is now content/common/cursors/webcursor.h

It looks like those are the only changes

Hi. I can't find a definitive answer to your question about UpdateCursor()
and non-desktop platforms. Browsing the code doesn't reveal any special
casing, but maybe I've missed something.

If you're concerned about the overhead; I've made a change to store the
FormFactor in a static variable, it should cause no problems in this
circumstance and avoid overhead on subsequent calls.

I've also made the changes to match changes in upstream.

I'm not particularly concerned about the overhead - just whether its necessary (neither qtwebengine or qtwebkit seem to do it) or correct (ie, is the form factor a reliable indicator of not having a cursor input)?

I'll test it on my Nexus 4 to see what happens on !desktop

I didn't get a chance to try this yet - there's quite a few build failures against the latest Chromium. Whilst WebCursor moved to from webkit to content, blink::WebCursorInfo hasn't changed

I've fixed up the build issues and merged this now. Thanks!