It is now possible to scale the current layer using the move tool from
all four corners of the layer boundary. Previously scaling was only
possible from the bottom right of the image.
This is required for me to be able to build both Serenity and
Ladybird from the same repo. Without this the two builds seem to
stomp on each other, then fail to link.
Repeatedly allocation a new Card object is unnecessary, and makes
propagating OOM awkward. We also don't need a full card, just which
suit/rank it is and its position. So, let's save all the extra
allocation and just paint the card bitmap directly.
For example, in Solitaire, the vertical normal stacks cover the suit of
all but the topmost card in the stack. To see the suit of covered cards
the user currently has to move the cards on top of them out of the way.
This adds an API for games to set a card at a location to be previewed,
which will draw that card on top of all other cards without moving it.
This patch also updates corresponding functions from
`LibFileSystemAccessServerClient`.
From the FileSystemAccessClient point of view, it only makes the server
take `Core::Stream::OpenMode` instead of `Core::OpenMode`. So, `enum`
conversions only happen within deprecated functions and not in the new
`Core::Stream` friendly API.
On the server side, it just removes two usages of `Core::File::open()`.
Generic PR actions include opening a PR, submit review comments, adding
new commits, etc. This prevents the reviewer and PR submitter from
having to manually bounce the labels back and forth in the general
case. The reviewer also may not have permission to set labels, meaning
the reviewer won't be able to update the labels accordingly themselves.
This does not handle more subjective labels such as pr-is-blocked and
pr-unclear. Unfortunately, there does not seem to be a GitHub Actions
trigger for when a PR has merge conflicts, so the pr-has-conflicts
label cannot be automatically applied.
Co-authored-by: kleines Filmröllchen <filmroellchen@serenityos.org>
No need to use a TextLayout here, we can just count the number of lines
and multiply that by the font's preferred line height.
In addition to being much simpler, it also fixes a bug where labels were
got too tall if we calculated their preferred height before assigning
a final width to them.
This is achieved by simplifying the logic in TextLayout. We get rid
of all the various ways that the layout bounding rect can get cropped.
Then we make sure to use the right pixel metrics.
Finally we use the font's own line gap metrics instead of hard-coding 4.
The end result is that text painted with vector fonts now gets pretty
reasonable vertical alignment in most cases.
The height of a line or column doesn't change unless the font changes,
and we were already caching the line height. This patch extends it so
we also cache the column width.
When changing the font size, we now resize the terminal widget *before*
setting the font. This ensures that we keep the same logical terminal
size after the font change.
Instead of indicating which individual cards should be highlighted, card
games now indicate which stack is highlighted. This lets the stack draw
empty stacks with a highlight (e.g. the Foundation stack in Solitaire).
If the stack is non-empty, the stack can delegate highlighting to the
top-most card.
Currently, the outside of the card highlight has rounded corners, but
the inside has square corners. It looks a bit more polished if they are
both rounded.
Change column distribution to take in account is_length() and
is_percentage() width values instead of treating all cells like
they have auto width by implementing it in the way described
in CSS Tables 3 spec:
https://www.w3.org/TR/css-tables-3/#width-distribution-algorithm
distribute_width_to_column() is structured to follow schema:
w3.org/TR/css-tables-3/images/CSS-Tables-Column-Width-Assignment.svg
It is not possible to use width of containing block to resolve
cells width because by the time compute_table_measures() is
called row width is not known yet.
This seems to work perfectly OK on my ICH7 test machine and also it
works on QEMU, so it is probably OK to restore this.
This will ensure we always get scan code set 1 input, because we enable
scan code set 2 and PS/2 translation on the first (keyboard) port.
The setting of scan code set sequence is removed, as it's buggy and
could lead the controller to fail immediately when doing self-test
afterwards. We will restore it when we understand how to do so safely.
Allow the user to determine a preferred detection path with a new kernel
command line argument. The defualt option is to check i8042 presence
with an ACPI check and if necessary - an "aggressive" test to determine
i8042 existence in the system.
Also, keep the i8042 controller pointer on the stack, so don't assign
m_i8042_controller member pointer if it does not exist.
