absolutely-positioned flows separately when storing overflow.
This code dates back to the time when absolutely positioned flows were
ignored by all of their ancestors up to the containing block. This
hasn't been true for at least a year.
Closes#9306.
Closes#9309.
Is a partial fix for #9308.
Also change when pipelines become active.
This makes the constellation activate a pipeline as the current frame
when it is ready to do initial reflow, rather than when it is ready
to paint.
This fixes a number of intermittent failures that could previously occur
if an iframe was not visible - which would mean it was never moved from
a pending frame in the constellation to an active frame.
(It happens that webrender exposes these intermittents as permanent failures).
Move Stylesheet loading and ownership from the layout task into HTML elements
Stylesheets for `HTMLLinkElement`s are now loaded by the resource task, triggered by the element in question. Stylesheets are owned by the elements they're associated with, which can be `HTMLStyleElement`, `HTMLLinkElement`, and `HTMLMetaElement` (for `<meta name="viewport">).
Additionally, the quirks mode stylesheet (just as the user and user agent stylesheets a couple of commits ago), is implemented as a lazy static, loaded once per process and shared between all documents.
This all has various nice consequences:
- Stylesheet loading becomes a non-blocking operation.
- Stylesheets are removed when the element they're associated with is removed from the document.
- It'll be possible to implement the CSSOM APIs that require direct access to the stylesheets (i.e., ~ all of them).
- Various subtle correctness issues are fixed.
One piece of interesting follow-up work would be to move parsing of external stylesheets to the resource task, too. Right now, it happens in the link element once loading is complete, so blocks the script task. Moving it to the resource task would probably be fairly straight-forward as it doesn't require access to any external state.
Depends on #7979 because without that loading stylesheets asynchronously breaks lots of content.
<!-- Reviewable:start -->
[<img src="https://reviewable.io/review_button.png" height=40 alt="Review on Reviewable"/>](https://reviewable.io/reviews/servo/servo/8039)
<!-- Reviewable:end -->
Stylesheets for `HTMLLinkElement`s are now loaded by the resource task, triggered by the element in question. Stylesheets are owned by the elements they're associated with, which can be `HTMLStyleElement`, `HTMLLinkElement`, and `HTMLMetaElement` (for `<meta name="viewport">).
Additionally, the quirks mode stylesheet (just as the user and user agent stylesheets a couple of commits ago), is implemented as a lazy static, loaded once per process and shared between all documents.
This all has various nice consequences:
- Stylesheet loading becomes a non-blocking operation.
- Stylesheets are removed when the element they're associated with is removed from the document.
- It'll be possible to implement the CSSOM APIs that require direct access to the stylesheets (i.e., ~ all of them).
- Various subtle correctness issues are fixed.
One piece of interesting follow-up work would be to move parsing of external stylesheets to the resource task, too. Right now, it happens in the link element once loading is complete, so blocks the script task. Moving it to the resource task would probably be fairly straight-forward as it doesn't require access to any external state.
Sometimes positioned content needs to be layered on top of stacking
contexts. The layer synthesis code can do this, but the current design
prevents it because stacking contexts are stored in a separate struct
member. In order to preserve tree order, mix stacking contexts into the
positioned content list, by adding a new StackingContextClass
DisplayItem. Such items do not have a base DisplayItem.
In some ways this simplifies the code, because we no longer have to
have a separate code path in the StackingContextLayerCreator.
Fixes#7779.
Fixes#7983.
Fixes#8122.
Fixes#8310.
There is no easy way to pass position-relative-035 & also
pass both tests/ref/border_rounding_1px_invisible_issue_7184 and
tests/ref/text_decoration_underline_subpx.
https://github.com/servo/servo/pull/7161, "Snap rectangles to nearest
pixels...", did fix position-relative-035 (except
for OS X) but broke 1px borders.
layout: Centralize the logic that determines whether fragments get layers in the fragment.
…so that it can be activated when we're forcing
the creation of extra layers due to positioned descendants that
themselves have layers.
The newly failing tests were tests that accidentally passed due to
incorrect stacking order.
Closes#7281.
r? @mbrubeck
<!-- Reviewable:start -->
[<img src="https://reviewable.io/review_button.png" height=40 alt="Review on Reviewable"/>](https://reviewable.io/reviews/servo/servo/7291)
<!-- Reviewable:end -->
layers in the fragment, so that it can be activated when we're forcing
the creation of extra layers due to positioned descendants that
themselves have layers.
The newly failing tests were tests that accidentally passed due to
incorrect stacking order.
Closes#7281.
There is no need to compute next_collapsed_borders_in_block_direction for all
kind of flows when it's used only in the table row case. That also avoids a panic
when the next child is a table colgroup (this should not happen when iterating
over a table row).
into account only once.
There were two bugs here: (1) relative position applied to
scanned/unscanned text fragments independently of the container element
that applied that relative position, causing double-counting; (2)
relative position applied to inline block fragments independently of the
wrapped block itself, causing double-counting.
This commit also removes the `cascade_anonymous` function and the
related `Fragment` constructor. They were unused, and their
functionality has been replaced by the `modify_style_for_*` series of
functions.
Closes#7067.
