m_rotated_rectangle_path was unused and m_intersection_edge_path was
cleared/free'd each time it was used. So sticking in the class just
bloats the size.
Compared to traditional modal search, incremental search begins
matching as soon as the user starts typing, highlighting results
immediately. This refactors Itamar's work for HackStudio into a
common LibGUI widget to be used in all multi-line TextEditors.
Banners are abstract widgets which can house additional controls
and information on a temporary basis, popping in from the top of
their parent when needed.
When starting to drag the cursor below the text, we would start the
selection from the closest point in the line above in the last
fragment. This is not the behavior seen in other browsers, and it
causes weird behavior when the cursor is to the left of the last
fragment, for instances when trying to select with the text
`before <span>middle</span> after`.
By starting the selection from the _end_ of the last fragment,
including the line end character, selection behavior is similar to
that of other browsers.
This gives a slightly more reasonable result when comparing different
colors with low alpha values.
For example comparing white with alpha 100 to transparent:
Before
distance: 0.78
After
distance: 0.07
(Where distance is between 0 and 1)
The result is unchanged for comparing colors without alpha values.
Employ the same hardening that glibc and the Linux kernel use for
generating stack guards: zero the first byte of the guard such that
if C-style string functions read out of bounds on the stack, we do
not overwrite or potentially leak the stack guard.
We used size_t, which is a type that is guarenteed to be large
enough to hold an array index, but uintptr_t is designed to be used
to hold pointer values, which is the case of stack guards.
The main fix is stopping fill_path() scanlines being rounded to the
wrong pixel. The y-value is now floored to ensure it always snaps
in the same direction.
This was state only used by the parser to output an error with
appropriate location. This shrinks the size of ObjectExpression from
120 bytes down to just 56. This saves roughly 2.5 MiB when loading
twitter.
When errors are encountered by PlaybackManager, it attempts to switch
states to either Stopped or Corrupted. However, that causes it to set
the last presentation media time to the current playback time while the
last presentation time is unexpectedly negative because the seek never
ended.
Ending the seek before the state changes to Stopped or Corrupted
prevents this situation from happening.
Using source offsets directly means we don't have to synthesize any
SourceRanges and circumvents that SourceRange returns invalid values for
the last range of a file.
This issue could be triggered by evaluating a module twice. If it failed
during evaluation the cycle root was never set but that status is
evaluated. This failed in step 3 of Evaluate which assumes it has a
cycle root. This is a spec issue see:
https://github.com/tc39/ecma262/issues/2823
To fix this we check if the module itself has a top level capability
first before going to the cycle root.
Co-authored-by: Jamie Mansfield <jmansfield@cadixdev.org>
