Behavior-wise this only removes the HasAttr(src) check, and adds the IsEmpty()
check to the alt attribute value, since this function is only called for <img>
and <input>.
But it also cleans up a bit.
Differential Revision: https://phabricator.services.mozilla.com/D11194
Behavior-wise this only removes the HasAttr(src) check, and adds the IsEmpty()
check to the alt attribute value, since this function is only called for <img>
and <input>.
But it also cleans up a bit.
Differential Revision: https://phabricator.services.mozilla.com/D11194
This patch allows us to intercept invalidation requests for display
items, and avoid regenerating the display list for animated images which
are using SharedSurfacesAnimation.
Differential Revision: https://phabricator.services.mozilla.com/D7504
If we do not pass the high quality scaling flag than the resulting surface will be marked as cannot substitute, which is not accurate, so we don't want.
The only place that actually tries to be smart about the size is nsImageFrame::MaybeDecodeForPredictedSize. All other cases just ask for the intrinsic size.
The two most likely cases are that there are no decoded copies of the image, or there is one decoded (or in progress) copy of the image.
In the first case we will request decode at the instrinsic size, and then if we draw at a different size that draw will request the proper size. This doesn't change with this patch.
In the second case there is a decoded copy already available, this is likely from a draw call on the image, and that is the surface size that we want. So we save a decode. If we are actually drawing the image at two different sizes the second size will be slightly delayed, but we have the wrongly sized copy of the image that we can draw until then. This seems like a good tradeoff to avoid always decoding an instrinic size copy of images.
When generating display lists for WebRender, we were not caching the
draw result via nsDisplayItemGenericImageGeometry::UpdateDrawResult (or
similar) after completing CreateWebRenderCommands. This is important
because reftests use this to force sync decoding for images; it may be a
reason for image-related intermittent failures on *-qr builds.
Additionally, we may have been requesting fallback in cases where fallback
could not do anything more than WebRender could. For example, if we can't
get an image container yet, there is no point in requesting fallback
because it might just be we haven't started decoding yet. We should just
return the actual draw result in such cases.
When the current image for an nsImageFrame/nsDisplayImage is not yet
ready, we display the previous image, if any available, to avoid
flickering while we wait for decoding to finish. On the WebRender path,
this functionality was lost since we did not have the draw result
information with image containers. With the API updated in part 1, we
can now do this to avoid flickering.
In addition to the image container, the draw result can also be useful
for callers to know whether or not the surface(s) in the container are
fully decoded or not. This is used in subsequent parts to avoid
flickering in some cases.
All classes deriving from nsIFrame that did not have any subclasses themselves
(at the time of writing this patch) have been marked with `final`.
Some other Layout classes have also been made final, but this was opportunistic
while working on nsIFrame subclasses, and is definitely not exhaustive, further
patches welcome; refer to bug 1332680.
Advantages of marking a class final include:
- Allowing the compiler to devirtualize some method calls (i.e., calling
virtual functions directly instead of going through the vtable),
- Indicating that the class is not currently subclassed,
- Preventing subclassing without being aware that this would remove the
finalization benefits of the parent class.
`final` does not signify that these classes should *never* be subclassed, this
is left for developers to decide.
Differential Revision: https://phabricator.services.mozilla.com/D5020
When the frame updates for an animated image, it will trigger
nsImageBoxFrame::OnFrameUpdate to be called. We did not change this for
WebRender, and thus it was missing similar checks added to
nsImageFrame::InvalidateSelf as originally added in bug 1382985. This
caused us to ignore the frame update, and thus the animation never
appeared to progress.
Using references helps to see when stuff can and cannot be null.
I removed useless aTag / aNamespaceId arguments which are useless now that XBL
can't override them (bug 1450617), so FindXULData is the only one that keeps
them alive.
Also, I took the liberty of renaming a few fooComputedStyle variables to just
fooStyle, and clarify naming in some pseudo-element-related functions to say
originating element (the spec term) and avoid confusing it with the generated
_moz_generated_content_before / _moz_generated_content_after element.
Note that this is a partial state, more stuff will come in the future.
Differential Revision: https://phabricator.services.mozilla.com/D2326
MozReview-Commit-ID: 39B30doREUH
This way we reuse the same machinery everywhere for the content property.
The only difference is that we need to look at the parent style for content
instead of just our style, and at a given index.
Again, this is fine because changing content reframes, so no chance to mess up.
This allows the generated content stuff to not implement nsImageLoadingContent
and all that stuff, nor deal with events, which makes it much simpler IMO.
Now it just tracks an index. We may not even need for it to be an HTML element,
but I've kept that for now.
I added a crashtest that used to crash because of the bogus
nsCSSFrameConstructor code which trusted the node name without checking it was
native anonymous.
Differential Revision: https://phabricator.services.mozilla.com/D1897
MozReview-Commit-ID: 1pAzIvRRVnL
This way we reuse the same machinery everywhere for the content property.
The only difference is that we need to look at the parent style for content
instead of just our style, and at a given index.
Again, this is fine because changing content reframes, so no chance to mess up.
This allows the generated content stuff to not implement nsImageLoadingContent
and all that stuff, nor deal with events, which makes it much simpler IMO.
Now it just tracks an index. We may not even need for it to be an HTML element,
but I've kept that for now.
I added a crashtest that used to crash because of the bogus
nsCSSFrameConstructor code which trusted the node name without checking it was
native anonymous.
Differential Revision: https://phabricator.services.mozilla.com/D1897
MozReview-Commit-ID: 1pAzIvRRVnL
I thought of just moving the tracking to nsImageFrame instead of
nsImageLoadingContent entirely, though that would mean I need to handle it also
in nsImageBoxFrame and nsSVGImageFrame, which was even more duplicated code.
Ideas for testing this welcome, though all our animated image test-cases are
borked (all reftests in image/test/reftest/apng are disabled, and all the ones
for gifs that have animations as well).
I'll cross-ref this bug and bug 1473651 so that we can write a test for this
once that's fixed.
Differential Revision: https://phabricator.services.mozilla.com/D1994
Dynamic changes are handled correctly because content property changes already
cause a reframe.
This implements the same bits as Blink / WebKit do (single content item which is
an image, otherwise gets ignored), except for the edge cases where you use this
on an image.
In order to handle the edge cases right, we completely isolate the
nsImageLoadingContent usage based on `mKind`.
Blink's and WebKit's behavior there makes no sense and it's erratic, what I
implemented is consistent (we apply to images as long as they don't generate a
box, and we don't look at alt text or broken icons), though I'll update to
whatever the WG decides in https://github.com/w3c/csswg-drafts/issues/2831 /
https://github.com/w3c/csswg-drafts/issues/2832.
I don't think it matters in terms of web compat in any case.
MozReview-Commit-ID: JUurhC60hWr
When computing image to layout transforms for invalidation, use the actual
intrinsic size of the source image, instead of the layout intrinsic size, which
may be scaled by ResponsiveImageSelector since bug 1149357.
I have absolutely no idea how to write a test for this, suggestions welcome.
MozReview-Commit-ID: LP6C9fSvMi2
GCC doesn't like StyleComplexColor with constructor in an anonymous
struct in an anonymous union. Replace the use of a union to access
`mBorder[..]Color` fields as an array with an accessor methods.
MozReview-Commit-ID: 1Wulh1qKYCZ
The methods where it's used don't run from reflow (they're image notifications
that run off runnables and such), so should be an idempotent change.
MozReview-Commit-ID: LdmSOcKDdw1
The frame is notified via its mListener, which is an observer of the
nsImageLoadingContent (mContent).
This last one only notifies for the current and pending requests, otherwise it's
a bug we need to fix there, not wallpaper here, since that'd mean that we forgot
to cancel the previous request. Added assertions to that effect.
Notify() is only called with the this object as a first argument from
imgRequestProxy, so it'd better be non-null, too.
MozReview-Commit-ID: DHaOLph2EAo
So that the imageLoader code is all grouped together. LoadIcons should probably
be a static, and the change shouldn't have any observable behavior.
MozReview-Commit-ID: Anxe8c5OfLe
Before that we were not notifying the image frame in any way if we ended up not
doing a load, and we were instead relying on the reflow the viewport resize
caused to get the new density in ComputeSize from the content node (but nowhere
else, since that's the bug part 1 fixes).
This was generally unsound, since you can stash random media in a sizes=
attribute, which don't necessarily needs to cause a reflow.
Now we need to notify necessarily because nsImageFrame stores the adjusted
intrinsic size.
mCurrentDensity could also get out of sync as well, when the selected image
density changed, but we ended up returning early because our source hadn't
change in the first early exit.
This patch moves us to a model where we don't re-trigger loads for density
changes if the source doesn't change (unless we pass aAlwaysLoad when we need
to, per spec).
This matches our previous behavior (without the bugginess of not updating the
intrinsic size), and also Chromium, at least.
This changes behavior in one case, which is when we don't load the same source
node, but we have the same source URL, and the density does change. This could
happen with <picture> and two <source>s with same source and different media and
sizes. This makes our behavior consistent with the behavior we have when both
the source and the density doesn't change.
Blink and WebKit do trigger a second image load both when the source changes
without changing density and when density changes. I'll file a spec issue, since
per:
https://html.spec.whatwg.org/#reacting-to-environment-changes
We should be triggering the load when the density changes but the source
doesn't as well, but no UA does that.
I filed https://github.com/whatwg/html/issues/3709 with a little summary of the
situation and what I think the behavior should be (which is what this patch
implements). That being said, I'll update the impl if the spec people think
otherwise :).
MozReview-Commit-ID: Eqy16ygHRLo
Only doing it in ComputeSize (via GetNaturalSize) is unsound, and the rest of
the users of mIntrinsicSize definitely do need scaling accounted for.
Move the adjustment to nsImageFrame for two reasons:
* Prevents adding more dependencies from nsIImageLoadingContent, which
otherwise would need to go away anyway in bug 215083.
* Avoids having to duplicate the image orientation logic, since mImage is
already an OrientedImage if needed.
MozReview-Commit-ID: EA0n0TctZhN
