Paul Berry <[email protected]> writes: > When running piglit in manual mode (no "-auto" command-line option is > given), there's no point in calling piglit_display() before the window > is exposed. This just leads to confusion because it will be called > again once the window appears, leading to duplicate test output in the > console window. > > When running piglit in "-auto" mode, if the test accesses the front > buffer, then we need to wait until the window is exposed before > calling piglit_display(), since a front buffer is not guaranteed to be > available before then (in practice, this only happens with > non-compositing window managers). However, if the test doesn't access > the front buffer, then it is still safe to call piglit_display() > before the expose event; doing so avoids wasting time.
I've been trying to decide whether I prefer "be a little more efficient and skip waiting for the event when we don't need it" like this commit, or "always wait for the event for sanity". Since I don't seem to have a strong preference, patches 1 and 2 are: Reveiwed-by: Eric Anholt <[email protected]>
pgpP3aMqOHzS3.pgp
Description: PGP signature
_______________________________________________ Piglit mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/piglit
