On 10/04/2012 01:38 PM, Eric Anholt wrote: > Chad Versace <chad.vers...@linux.intel.com> writes: > >> I would have used the checked variant of the xcb call, >> xcb_dri2_wait_xbc, and then returned 1 if an error had occured. Why >> did you choose to use the unchecked variant? >> >> I'm aware that the old DRI2WaitSBC also always blindly returned >> success, but is there justification for the new code to also do that? > > Preserves current behavior of passing errors through. The spec is a > badly done port of the WGL version and specifies: > > 1) Returning a boolean error value for any error in the wait functions. > 2) Integer error values to be returned for various cases. > > Those integer error values are specified from the set returned by > glXGetConfig(), not the set of GLX errors used by everything else in > GLX. > > Our conclusion was that what the spec author would have meant was that > those errors specified mean "pass these errors through, except using > actual GLX error names", and the weird bool error return from these > functions would be for whatever other sort of error might occur in > processing.
Ok, thanks for the explanation. For patches 11-14, Reviewed-by: Chad Versace <chad.vers...@linux.intel.com> _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
