Re: ARGB cursors (was Re: Adjusting the 2.4 schedule)

[Keith Packard <keithp keithp com>]

Around 22 o'clock on Jun 26, Matthias Clasen wrote:

> - Is the size returned by XQueryBestCursor actually binding for ARGB
>   cursors created via the Render extension ?

It happens to be informative for known hardware, so perhaps we should make 
that notation in the spec.  However, ARGB cursors are implemented so that 
a software fallback always presents the entire cursor when possible.  The 
only time the software fallback isn't usable is when a DRI or Xv window is 
viewable -- in that case, the X server should force a transition back to 
the hardware cursor to avoid possible screen corruption.  For hardware 
without ARGB cursor support, that means switching back to two color 
cursors.

> - Are the Xcursor query functions sufficient ? They only take the    
>   display as an argument, so if there is one color and one monochrome
>   screen, will XcursorSupportsARGB return true or false ?

X cursors are screen independent; XcursorSupportsARGB returns only whether 
the X server has the appropriate version of the Render extension.  Of 
course, the presentation of the cursor image will depend on the underlying 
hardware (or software) support in the server, so it's likely (but not 
specified) that a monochrome screen will present only a monochrome cursor 
image.  Hence the desire for a 'default' cursor theme which presents 
clearly on monochrome screens.

> Any comments on the idea to supply cursors with a name (*) for a11y ?

Without a spec for the names, I'm not sure this isn't worse than not 
including it.  But, perhaps a tentative spec could be worked out 
contemporaneously with the propogation of the new API into applications.

-keith