Code review comment for lp:~kdub/mir/display-groups
Revision history for this message
| Daniel van Vugt (vanvugt) wrote | # |
© 2004
Canonical Ltd.
•
Terms of use
•
Data privacy
•
Contact Launchpad Support
•
Blog
•
Careers
•
System status
•
bbfa235
(Get the code!)
> > I can only imagine this works at all on Android without tearing or skipping
> > (does it?) if Android is abstracting and asynchronously deferring the page
> > flipping in the background.
>
> Yes, the kernel helps out to avoid tearing. We're guaranteed that the set()
> function won't block, so its more that we're scheduling work with the driver
> and the kernel than controlling the timing of the threads.
Right, so this works for Android because Android isn't a low-level display system. It's a high-level abstraction of displays (and very much OpenGL-only). Native platforms (mesa, nvidia etc) will usually be low-level display systems that need asynchronous posting (hence never use DisplayGroup).
Although I can now imagine some virtual Mir platform in a window. If you wanted to represent multiple virtual outputs for visualization all in a single window then DisplayGroups would be used. But I doubt any such use case beyond Android will really ever be required.