0672055a72ff1265b2aedb037f2848455318f22a changed the Material widget to
always use Paths for representing the outline.
These paths are later used for clipping and drawing a shadow.
This changed introduced a performance regression, see:
https://github.com/flutter/flutter/issues/14403
We did not expect a path that is a rounded rectangle to be less
performant than a rounded rectangle, as Skia should be able to tell the
path is just a rounded rectangle.
Until we find a solution for this regression, we keep using RRect when
we can represent the shape with it.
For backward compatibility we keep supporting specifying the shape as a
combination of MaterialType and borderRadius, and we just use that as a
default when shapeBorder is null.
To cleanup the implementation if shapeBorder was not specified we just
translate the specified shape to a shapeBorder internally.
I benchmarked paint, layout and hit testing, with the specialized shape
clippers vs. the equivalent path clippers and did not see any
significant performance difference.
For testing, I extended the clippers/physicalShape matchers to match either the
specialized shape or the equivalent shape.
Seems like we don't yet run the flutter tests on Windows, but we're
close to being able to. This makes some minor changes to make that
more possible:
- fix the stack parsing code to support Windows paths
- fix the tests for the stack error handling code to handle Windows paths
- skip some tests that rely on Ahem font metrics
* libtxt now calculates selection rectangles reflecting the heights of
differently styled text spans within a line (matching Blink)
* libtxt does not round the coordinates of selection rectangles (unlike Blink)
Prevent header from thinking it can wrap and then overflowing.
Fix default footer string which lost its colon (localized values are fine).
Make the "rows per page" drop-down include at least one value even when the table lacks many items. (Previously it would assert if your table was too short.)
Make the footer scrollable.
Fix some todos and improve some debug output.
Tests for much of the above.
* make time picker accessible
* use new CustomPaint a11y API
* flutter_localizations tests; use bigger distance delta
* fix am/pm control; selected values
* fix translations; remove @mustCallSuper in describeSemanticsConfiguration
* exclude AM/PM announcement from iOS as on iOS the label is read back automatically
This allows the scheduler library to depend on the services library
and the painting library to depend on the scheduler library without
the services library having to depend on the scheduler library.
While I was at it I also cleaned up some of the binding logic: the
licenses logic can now be overridden (and the test library does so),
and the image cache can now be overridden as well.
* ensures that only semantics boundaries will be added to owner._nodesNeedingSemantics as expected by compiler.
* no longer throws assert if markNeedsSemanticsUpdate is called on non-semantic-boundary render object with a non-semantic-boundary parent.
* Fixes#13109.
* removes onlyLocalUpdates from markNeedsSemanticsUpdate as its no longer needed.
This fixes the popup menu code to do a better job of expanding
smoothly regardless of which side of the screen it's on. It still
results in a bidirection growth when positioned at the bottom of the
screen, so maybe we'll need to animate menus differently, but that's
a problem for another patch.
Also, improve some docs and provide RelativeRect.toSize which I needed
at one point while building this patch (though it didn't survive all
the way to the end).
I got tired of drive-by spelling fixes, so I figured I'd just take care of them all at once.
This only corrects errors in the dartdocs, not regular comments, and I skipped any sample code in the dartdocs. It doesn't touch any identifiers in the dartdocs either. No code changes, just comments.
This makes ListTile expand vertically when text is scaled, or really when whatever is placed inside it is larger than the available space.
In order for UnconstrainedBox to be useful here, I needed for it to only unconstrain the child Row in one dimension, so I added a "constrainedAxis" parameter to the UnconstrainedBox.
Also, changed one enum test to use a switch instead.
I modified the ListTile test to be more representative of the intention of the spec: we were testing with text in the leading and trailing sections, and the design wants icons there. Because there was leading text, and the dense mode only changes the font size on the text lines, the leading text was propping up the minimum size of the tile, making so that the test wasn't really testing any changes in dense mode.
UnconstrainedBox will allow its child to size itself as if it had no constraints, and then attempt to fit around that object, until its own constraints are exceeded, in which case it will clip and display an overflow warning.
I also factored out DebugOverflowIndicator, which will draw overflow indicators on containers which overflow but aren't expected to.
libtxt's implementation of Paragraph.getBoxesForRange returns the full line
height for all characters in a line. Blink had been returning a tighter
enclosing rectangle for each individual character.
Also:
* Provide a better message when you lerp from infinity to finity
constraints.
* Make the striped marker support RTL.
* By popular demand, dump a warning to the console the first time
a particular Flex overflows. (Resets on hot reload.)
