These were totally ad-hoc before, is_inline() was based on a boolean
flag on Layout::Node that we set in various situations.
Meanwhile, is_inline_block() was a combination on is_inline() plus a
type check to see if the layout node inherited from BlockContainer.
This patch replaces the above mess with simple lookups of the CSS
display value. Note that layout nodes without their own style (i.e text
nodes) are automatically assumed to be inline and non-blocks. This has
to be special-cased since layout nodes without style will consult the
style of their parent, so without short-circuiting this would break.
This is one of many small steps towards being able to remove the ad-hoc
Layout::Node::is_inline() in favor of honoring the CSS display value
everywhere instead.
Until now, VERIFY() failures would just cause a __builtin_trap() in
release builds, which made them a bit too harsh. This commit adds an
out-of-line helper function that prints the error before trapping.
We always create a Layout::InitialContainingBlock for the ICB, but in a
future where we always honor the CSS::Display everywhere, we need to
make sure everyone has the right display values.
The mappings are exposed via `Unicode::code_point_decomposition(u32)`
and `Unicode::code_point_decompositions()`, the latter being useful for
reverse searching a code point from its decomposition.
The normalization code does not make use of `Quick_Check` props (https://www.unicode.org/reports/tr44/#Decompositions_and_Normalization),
meaning no quick check optimizations.
Currently the default is false, but this is not the best strategy
for most filters producing artifacts on the borders, and wrap-around
option ends up being better, producing less artifacts.
This prevents items from being put in the overflow menu, even though
there is still enough space for all items to be shown, because the
overflow button does not take up space when it is not needed.
By making proper use of the dynamic layout system, the Breadcrumbbar can
now shrink below its current (grown) size again, while still ensuring
that no icon gets cut off.
When items are sent to the overflow menu one by one, it can happen that
buttons that are heavily related, and don't make sense without one
another (either visually or logically) are separated.
This new option enables the developer to choose the "grouping"
behavior, of sending all items that are not separated to the overflow
menu together, as soon as one of them doesn't have enough space to be
displayed. (provided the toolbar is set as collapsible)
Previously if the user closed the last open tab in the last TabWidget
then the open file action would stop working until they explicitly
went to the View menu and added a new editor group.
Before this commit there was an attempt at preventing the user to
close the last open tab by hiding its close button, this didn't
account for the many other ways one could close it though.
Previously when trying to open a file from the tree view the file
would open in the currently selected tab, substituting the file
we were previously reading.
This change makes it so that clicking on a file from the tree view
opens it in a new tab, or selects the tab containing that file if
it's already open in the selected editor group.
This will allow Ladybird to use IPC::Connection without having an
actively running Core::EventLoop.
The abstraction here is not great, and we should think of something
nicer, but we have to start somewhere.
This patch adds WebView::ViewImplementation with all the
`notify_server_did_this_or_that()` functions from OOPWV.
This will allow us to share code between different web views, paving the
way for a Qt widget in Ladybird that can talk to a WebContent process.
Our IPC protocol currently relies on the behavior of recvfd() and
sendfd() on SerenityOS, which provide an out-of-band queue that can be
accessed independently of the in-band data stream.
To make LibIPC usable on other platforms, this patch adds a mechanism
where IPC::Connection can be given a dedicated socket for FD passing.
This gives us the same behavior as the syscalls on SerenityOS, without
having to change the protocol implementation.
With the addition of the 'fetch params' struct, the single ownership
model we had so far falls apart completely.
Additionally, this works nicely for FilteredResponse's internal response
instead of risking a dangling reference.
Replacing the public constructor with a create() function also found a
few instances of a Request being stack-allocated!
Doesn't use them in libc headers so that those don't have to pull in
AK/Platform.h.
AK_COMPILER_GCC is set _only_ for gcc, not for clang too. (__GNUC__ is
defined in clang builds as well.) Using AK_COMPILER_GCC simplifies
things some.
AK_COMPILER_CLANG isn't as much of a win, other than that it's
consistent with AK_COMPILER_GCC.
