Consider this testcase:
```html
<canvas style="aspect-ratio: 1; height: stretch; background: cyan"
width="200" height="100"></canvas>
```
To compute the intrinsic inline sizes we were treating `height: stretch`
as the natural height (100px) and then transferring that to the inline
axis through the preferred aspect ratio. So the element was 100px wide.
However, an indefinite `stretch` should be treated as an automatic size,
which wouldn't be transferred to the inline axis.
The fix actually makes the code slightly simpler.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
This allows `SameFormattingContextBlock` to cache inline content sizes
and will eventually allow it to participate in incremental layout.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
This allows cells to cache their inline content size and will eventually
allow them to participate in incremental layout.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Add a new struct `LayoutBoxBase`, that will be used throughout the box
tree. The idea of this struct is that we have a place to consistently
store common layout information (style and node information) and also to
cache layout results such as content sizes (inline and maybe later box
sizes) and eventually layout results.
In addition to the addition of this struct,
`IndependentFormattingContext` is flattened slightly so that it directly
holds the contents of both replaced and non-replaced elements.
This is only added to independent formatting contexts, but will later be
added to all block containers as well.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
We were sizing absolutely positioned replaced elements within their
actual containing block instead of the inset-modified containing block.
Then the `stretch` keyword would result in a wrong size.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
* Add missing support for some alignment keywords on absolutely positioned elements
Signed-off-by: taniishkaaa <tanishkasingh2004@gmail.com>
* Check the direction of the alignment container, nits, test expectations
In this case we need to check the direction of the static position
containing block, not the actual containing block:
```html
<!DOCTYPE html>
<div style="position: relative">
<div style="display: flex; flex-direction: column; width: 100px; height: 100px; border: solid; direction: rtl">
<div style="position: absolute; width: 20px; height: 20px; background: cyan; top: 20px; align-self: self-start"></div>
<div style="position: absolute; width: 20px; height: 20px; background: magenta; bottom: 20px; align-self: self-end"></div>
</div>
</div>
```
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
---------
Signed-off-by: taniishkaaa <tanishkasingh2004@gmail.com>
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
* Obey min and max properties when computing main size of column flex
When laying out a column flex container with an auto preferred main size,
we were resolving the used main size to the intrinsic max-content size.
However, we weren't clamping this amount between the min and max sizes.
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
* Improve performance of flex column layouts by caching
We were already using a cache for layout_for_block_content_size(), but
we were only storing the intrinsic block size. Thus when laying out the
flex items for real, we would perform new layouts, triggering an
exponential complexity in case of nested flexboxes.
Now we cache the entire layout result so that we can avoid doing the
work again.
This improves the results of flexbox-deeply-nested-column-flow.html
(a Blink perf test) from ~40 runs/second to ~500 runs/second on my PC.
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
---------
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
When laying out a column flex container with an auto preferred main size,
we were resolving the used main size to the intrinsic max-content size.
However, we weren't clamping this amount between the min and max sizes.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
`AbsoluteAxisSolver::solve()` would compute, among other things, the
position of the absolute positioned element if it had start alignment.
Then, `AbsoluteAxisSolver::origin_for_alignment_or_justification()`
could optionally opt into modifying that alignment if needed.
This was quite convoluted and not easy to follow. It's simpler to not
compute the position in `AbsoluteAxisSolver::solve()`, and instead do it
always in `AbsoluteAxisSolver::origin_for_alignment_or_justification()`,
which I'm renaming to `AbsoluteAxisSolver::origin_for_margin_box()`
because it aligns the margin box of the abspos within its alignment
container.
Then the `Anchor` struct becomes useless and can be removed.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
The containing block for the static position of an absolutely positioned
element in flex layout is established by the flex container. However, if
the flex container has static position, the actual containing block will
be established by another ancestor.
If the flex container and the containing block have different directions,
the static position needs especial handling when aligning the abspos.
We were already trying to do so with the `flip_anchor` flag, but there
were bugs.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
* Refactor computation of preferred aspect ratios
Computing min/max-content sizes required a ContainingBlock in order to
resolve the padding and border when determining the preferred aspect
ratio. However, all callers already knew the padding and border, so they
can compute the ratio themselves, and pass it directly instead of
the ContainingBlock.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
* Put preferred aspect ratio into ConstraintSpace
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
---------
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
We were using the preferred aspect ratio provided by the `aspect-ratio`
property instead of the natural aspect ratio. However, the preferred
aspect ratio should only be used to size the replaced element. To paint
the replaced contents into that element we need the natural ratio.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
- Clamp the stretch size to not be negative when the sum of padding,
borders and margins exceed the available space. This avoids a 2nd
layout.
- Avoid computing the inline content sizes if the result isn't needed.
- Instead of clamping both the min-content and max-content sizes to be
between the min and max constraints, just compute the fit-content size
first, and then clamp. Then `ContentSizes::map()` can be removed.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
* Remove unused deps
This doesn't seem to remove any deps from the workspace.
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
* ohos: Remove gaol dependency
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
---------
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
When building scroll frames, add a special
`StackingContextContent::Fragment` type for a hit test that covers all
scroll frame contents. This makes it so that you don't have to be
hovering over actual content to scroll the scroll frame.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Mukilan Thiyagarajan <mukilan@igalia.com>
If a flex item in a single-line column flex container stretches, then
we can know its final size. So instead of first laying it out using its
intrinsic inline size, and then stretching it later, we can use the
correct size from the very beginning.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
It was only used for the style, but the containing block of the caption
is the table wrapper box, whose style is stored in `self.table.style`.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
* Clean up tracing instrumentation
Signed-off-by: Delan Azabani <dazabani@igalia.com>
* Set all tracing spans to trace level for now
Signed-off-by: Delan Azabani <dazabani@igalia.com>
---------
Signed-off-by: Delan Azabani <dazabani@igalia.com>
The min-content and max-content sizes on the block axis depend on the
inline size. But when computing the SizeConstraint corresponding to the
inline axis, we were resolving the preferred inline size ignoring
intrinsic keywords. Now we will only ignore `auto`.
Also, this patch refactors the logic to compute the min-content and
max-content block sizes after fully resolving the inline size.
This avoids having to resolve the inline sizing properties twice.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
* Support justify-self on absolutely positioned elements
Signed-off-by: L Ashwin B <lashwinib@gmail.com>
* updating test expectations
Signed-off-by: L Ashwin B <lashwinib@gmail.com>
---------
Signed-off-by: L Ashwin B <lashwinib@gmail.com>
In order to support size keywords in block layout, we may need to call
`inline_content_sizes()` in order to compute the min/max-content sizes.
But this required a mutable reference in order the update the cache,
and in various places we already had mutable references.
So this switches the cache into a RwLock to avoid needing mutable refs.
Note OnceCell wouldn't work because it's not thread-safe.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
In most cases we already had a LazyCell anyways, since we could need the
value for multiple properties. Instead of passing a callback that forces
the evaluation of the LazyCell, it's simpler to just pass the LazyCell
directly.
Also, this way we no longer need mutable references.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
To compute the min-content and max-content inline sizes of a replaced
element, we were only using the aspect ratio to transfer definite block
sizes resulting from clamping the preferred block size between the min
and max block sizes.
However, if the preferred block size is indefinite, then we weren't
transfering the min and max through the aspect ratio.
This patch adds a `SizeConstraint` enum that can represent these cases,
and a `ConstraintSpace` struct analogous to `IndefiniteContainingBlock`
but with no inline size, and a `SizeConstraint` block size.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
When a flex item stretches in the cross axis in a row flex, the flex
container layout should depend on block constraints. In this case the
cross axis is the block axis (assuming horizontal writing modes --
vertical are not yet supported). This changes fixes an issue where the
cached layout was used in this case when stretching should trigger a new
layout.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
Instead of doing so much zipping, which is confusing, create a temporary
data structure for each item that holds all relevant information. In
addition, add detailed specification text so it is easier to understand
what is going on.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Previously, when there were no more violations, the returned value for
line free space was incorrect for flexible length resolution. It was
returning the container main space minus the inner length of each item.
Free space is determined by the outer length though. Fix this by reusing
the `free_space()` function, but with an argument indicating that all
items are now frozen.
Fixes#34079.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
This adds parallelism using rayon to the initial layout of flex lines
and line item. There is still no support for parallelism when laying out
line items again to account for stretch. This can be done in a followup
change.
In addition, the array of data for each line item is collected into a
new structure, `FlexLineItem`. This prevents a lot of needless zipping
of iterators at the expense of some new vector allocations. A folluwup
change can likely unify this structure and `FlexItemLayoutResult` though
that will require a larger refactor.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
We don't need to floor the preferred box sizes to be at least zero,
since `used_size_as_if_inline_element_from_content_box_sizes()` will
take care of applying min and max constraints, and the min has been
floored to be at least zero.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
The logic varied quite a bit depending on the case, now it's unified.
This also fixes the following case where the iframe was 150px tall
instead of 50px:
```html
<iframe style="min-width: 400px; max-height: 50px"></iframe>
```
This also modifies video-intrinsic-width-height.html to expect the new
behavior that we share with Blink and WebKit. In fact WebKit already
modified this test but forgot to export the change upstream. Firefox is
different but it was already failing anyways.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
A `<video>` element with no source won't have a natural aspect ratio,
but `aspect-ratio: auto` should still fall back to a ratio of 300/150.
`used_size_as_if_inline_element_from_content_box_sizes()` was already
handling this, but other consumers of `preferred_aspect_ratio()` were
wrong. In particular, this resulted in a 0px wide inline-block:
```html
<div style="display: inline-block; border: solid">
<video style="height: 100px; background: cyan"></video>
</div>
```
So this patch moves the fallback into `preferred_aspect_ratio()`.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
We were following CSS2, which didn't handle `aspect-ratio`.
This patch simplifies the logic and handles it correctly.
Unfortunately this makes 2 tests fail, but I'm pretty sure they aren't
spec-compliant. I'm leaving them as-is for now since they are part of
interop-2021, and Gecko, Blink and WebKit pass them (because of some
non-interoperable incorrect behaviors).
I'm adding a new test that is fully passed by Servo and WebKit.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
