Code review comment for lp:~mir-team/mir/cursor-spike-phase-2-resubmit

>> (1) Reading again, I still believe there's no justification for MirCursorConfiguration. All the relevant '
>> info would be contained in a MirCursor (image + hotspot... what else is there?).

The idea in changing it was that MirCursor could be meant to refer to a particular on-screen cursor. RAther what this struct contains is the parameters used to configure an on screen cursor.

>> (3) You should refer to them by identifier name in the docs.

Ok.

>> (4) OK, assuming the verb "configure" describes something more complex than "setting" a cursor, what's the extra complexity? What is it doing other than setting a cursor?

I am thinking of the cursor as the onscreen buffer with a particular location. In this sense you can't set a cursor on a surface, rather you set the cursor configuration which a surface requests.

>> (5) Your justification for having invisible cursors is not valid, I believe. Games with no visible cursor require relative movement data, which is unrelated to any "cursor" location

I guess that's true for racing games which would need some sort of mouse grab...I think there are still use cases though.

1. Virtual Machine/VNC type apps operating without mouse grab
2. Apps which may wish to use crazy cursors (i.e. maybe a fullscreen painting app has some crazy paint brush cursor which interacts with the surface in animations and its easiest to just do it on the client side).
3. Xmir (arguably, but its quite convenient!)