If a local include does not point to a file in the repository, it should
be a system include instead. This is now checked on every commit.
While this does introduce significant overhead in terms of percentage,
I think that an additional 10ms on huge commits (or less on smaller
commits) are acceptable:
hyperfine -w1 './Meta/check-style.py AK/*.h AK/*.cpp' # Before
Benchmark 1: ./Meta/check-style.py AK/*.h AK/*.cpp
Time (mean ± σ): 20.3 ms ± 0.4 ms [User: 17.1 ms, System: 3.
5 ms]
Range (min … max): 19.5 ms … 21.6 ms 128 runs
hyperfine -w1 './Meta/check-style.py AK/*.h AK/*.cpp' # After
Benchmark 1: ./Meta/check-style.py AK/*.h AK/*.cpp
Time (mean ± σ): 32.3 ms ± 0.4 ms [User: 27.9 ms, System: 4.
4 ms]
Range (min … max): 31.4 ms … 34.9 ms 91 runs
Instead of having a single available memory range that encompasses the
whole 0x00000000-0x3EFFFFFF range of physical memory, create a separate
reserved entry for the RAM range used by the VideoCore. This fixes a
crash that happens when we try to allocate physical pages in the GPU's
reserved range.
This will eventually be replaced with parsing the data from the device
tree, but for now, this should solve some of the recurring CI failures.
Depending on stack values being correctly and deterministically
overwritten was a bit too optimistic, to be honest. This new logic uses
a value on the heap.
This function generates a new path, which can be filled to rasterize
a stroke of the original path (at whatever thickness you like). It
does this by convolving a circular pen with the path, so right now
only supports round line caps.
Since filled paths now have good antialiasing, doing this results in
good stroked paths for "free". It also (for free) fixes stroked lines
with an opacity < 1, nice line joins, and is possible to fill with a
paint style (e.g. a gradient or an image).
Algorithm from: https://keithp.com/~keithp/talks/cairo2003.pdf
Dr. POSIX says:
Although the space used by string is no longer used once a new
string which defines name is passed to putenv(), if any thread in
the application has used getenv() to retrieve a pointer to this
variable, it should not be freed by calling free(). If the changed
environment variable is one known by the system (such as the locale
environment variables) the application should never free the buffer
used by earlier calls to putenv() for the same variable.
Applications _should_ not free the data passed to `putenv`, but they
_could_ in practice. I found that our Quake II port misbehaves in this
way, but does not crash on other platforms because glibc/musl `putenv`
does not assume that environment variables are correctly formatted.
The new behavior ignores environment variables without a '=' present,
and prevents excessively reading beyond the variable's name if the data
pointed to by the environment entry does not contain any null bytes.
With this change, our Quake II port no longer crashes when switching
from fullscreen to windowed mode.
Note that this fixes contradictory default values for group
"Preferences", key "Home". This is exactly the kind of errors I want to
prevent with this new style.
The hope is that this can later be used to:
- verify that all accesses to the same key use the same default value,
- and extract the default values more easily.
Like the HID, Audio and Storage subsystem, the Graphics subsystem (which
handles GPUs technically) exposes unix device files (typically in /dev).
To ensure consistency across the repository, move all related files to a
new directory under Kernel/Devices called "GPU".
Also remove the redundant "GPU" word from the VirtIO driver directory,
and the word "Graphics" from GraphicsManagement.{h,cpp} filenames.
The implemented cloning mechanism should be sound:
- If a PartitionTable is passed a File with
ShouldCloseFileDescriptor::Yes, then it will keep it alive until the
PartitionTable is destroyed.
- If a PartitionTable is passed a File with
ShouldCloseFileDescriptor::No, then the caller has to ensure that the
file descriptor remains alive.
If the caller is EBRPartitionTable, the same consideration holds.
If the caller is PartitionEditor::PartitionModel, this is satisfied by
keeping an OwnPtr<Core::File> around which is the originally opened
file.
Therefore, we never leak any fds, and never access a Core::File or fd
after destroying it.
This fixes an issue where images with padding and/or border did not have
their size adjusted for `border-box`, thereby becoming larger than
intended by the author.
This patch adds an alias to the source builtin when an user types ".".
We cannot just add an enumeration entry using __ENUMERATE_SHELL_BUILTIN
because "." is not a valid name in a function.
This patch adds handling similarly to the name rewriting of ":".
This alias is limited to POSIX mode only.
I was not aware of this framework back when implementing this back in
bc54560e59. Add in some basic tests for
this now that we are compliant with the specification.
This has KString, KBuffer, DoubleBuffer, KBufferBuilder, IOWindow,
UserOrKernelBuffer and ScopedCritical classes being moved to the
Kernel/Library subdirectory.
Also, move the panic and assertions handling code to that directory.
The "Crop Image to Selection" action will now crop all layers within
the current selection.
Any layers outside of the current selection are deleted.
If there are no layers within the current selection, a new layer the
same size as the selection is created, which has the same name as the
current active layer.
If a box has a negative margin-left, it may have a negative effective
offset within its parent BFC root coordinate system.
We can account for this when calculating the amount of left-side float
intrusion by flooring the X offset at 0.
