Removed aProtoDoc parameter from nsXULElement.h and nsXULElement.cpp in
content/xul/content/src. Specifically in method nsXULElement::Compile.
Removed aProtoDoc argument from XULDocument.cpp, and nsXULContentSink.cpp
in content/xul/document/src. Specifically when aProtoDoc argument used
in nsXULElement::Compile.
This method was originally exposed so that the animation compositor could test
if an animation should be skipped when sampling but now that we perform this
testing within the SVGAnimationElement itself it is no longer needed.
Remove the changes to nsSMILAnimationController introduced by bug 619469 now
that we have introduced a more thorough approach to disabling animations that
fail conditional processing tests.
Before the changes in this bug, this test used to produce an assertion:
ABORT: Rewind in the middle of a forwards seek?
This was because for a timed element whose corresponding animation element fails
conditional processing tests, we were still performing milestone samples even
though we were skipping regular samples.
This patch makes use of the SetIsDisabled method added to nsSMILTimedElement in
the previous patch to "turn off" the associated timed element of an animation
element that has failing conditional processing tests.
This method is added so that animation elements that fail conditional processing
tests can disable the corresponding timed element. Currently, this handling is
performed by the animation controller which simply skips sampling animations
that fail tests. However, this doesn't provide the correct behavior since
"skipped" animations can still have intervals that trigger syncbase dependencies
etc. Also, we still do milestone samples for "skipped" animations which can
lead to cases where we are in an inconsistent state.
This patch provides a method that will be used in subsequent patches to more
correctly remove the effects of a timed element.
To do this, this patch also factors out some common methods for clearing timing
state that can be shared with the Rewind method.
We're keeping the core idea that, before we remove the frames-to-be-destroyed
from the continuation chain, their textruns need to be disconnected/destroyed.
However, nsContinuingTextFrame::DestroyFrom tries to optimize when the
destroying frames that aren't mentioned in the userdata for the textrun, and
certain other conditions are met; we need a similar optimization here. It's
simpler here because the other conditions are definitely met, since all the
text for the frames being deleted has already been consumed and reflowed by
previous frames.
We don't need the TEXT_STYLE_MATCHES_PREV_CONTINUATION state bit anymore
because nsContinuingTextFrame::DestroyFrom will never see any textruns when
called via RemoveEmptyInFlows.
--HG--
extra : rebase_source : 6544f923499ef604d48ec15961716549dd25d279
