On Thu, Feb 06, 2020 at 02:20:02PM +0100, Gerd Hoffmann wrote:
On Thu, Feb 06, 2020 at 11:52:05AM +0000, Peter Maydell wrote:
> On Thu, 6 Feb 2020 at 11:29, Gerd Hoffmann <kraxel(a)redhat.com> wrote:
> >
> >
>
> This cover letter is missing a description of what the patchset does...
>
> The closest thing the patchset seems to get to documentation is the
> oneliner in ui.json:
> +# @show-cursor: Force showing the mouse cursor (default: off).
>
> but looking at the ui/cocoa.m implementation that isn't what it
> actually does -- it just seems to mean "default to shown on
> startup", because the logic that unconditionally hides the host
> cursor on mousegrab and unhides it on ungrab remains
> unchanged. This doesn't on the face of it sound like very
> useful behaviour, since the option will only have an effect for
> the short period of time between QEMU startup and the first
> mouse-grab, but without documentation of what the option
> is intended to do and in particular how it's intended to
> interact with grab/ungrab I don't know what your intention
> for the behaviour was.
Well, it doesn't change actual behavior for SDL and cocoa. It only adds
"-display {sdl,cocoa},show-cursor=on" as replacement for the global
"-show-cursor" option. Guess I should reorder the patches (move 5/6
before the individual UI patches) and reword the commit messages.
I suppose this patch set didn't intend to workaround the missing cursor over
VNC with NVIDIA vGPUs, right? IIRC the default is that the local cursor is
sent over to the destination whereas we need to force the remote cursor to be
sent back, do I remember that correctly the setting for which have to advertize
vncviewer?
Erik