Reverts recent change introduced to support implicit symbolic permission
which broke the parser when multiple classes are specified.
The state machine must assume it's dealing with classes until an
operation character is consumed.
This is a bit of a hack to get box content to stop bleeding 1px outside
the border sometimes. We will need to come up with a more general
solution for this problem eventually.
While IFC flows text into a block container, floating objects are
anchored at the BFC root, not necessarily the local block container.
Because of this, we have to use root-relative coordinates when checking
how much space is available in between left and right floated objects.
Block placement is now divided into a vertical and horizontal step. The
vertical step runs before formatting boxes internally. The horizontal
step still runs after (since we may need the final width value.)
This solves a long-standing architectural problem where IFC didn't know
its origin Y position within the BFC root box. This is required for
figuring out how to flow around floating objects. (Floating objects are
always relative to the BFC root.)
Rather than displaying the path of the framebuffer, try and display
the manufacturer name and the size of the display. If no EDID data is
available, fall back to showing the device path.
This downloads the UEFI's published PNP ID database and generates a
lookup table for use in LibEDID. The lookup table isn't optimized at
all, but this can be easily done at a later point if needed.
This library can be used (for the most part) by kernel drivers as well
as user mode. For this reason FixedPoint is used rather than floating
point, but kept to a minimum.
This patch reintroduces the translation previously mistakenly removed
when adding support for different underline-styles.
Thanks for reporting the bug, kennethmyhra!
The implementation of LIKE uses regexes under the hood, and this
implementation of REGEXP takes the same approach. It employs
PosixExtended from LibRegex with case insensitive and Unicode flags
set. The implementation of LIKE is based on SQLlite specs, but SQLlite
does not offer directions for a built-in regex functionality, so this
one uses LibRegex.
It's a bad idea to have a global event loop in a client application as
that will cause an initialization-order fiasco in ASAN. Therefore, LibC
now has a flag "s_global_initializers_ran" which is false until _entry
in crt0 runs, which in turn only gets called after all the global
initializers were actually executed. The EventLoop constructor checks
the flag and crashes the program if it is being called as a global
constructor. A note next to the VERIFY_NOT_REACHED() informs the
developer of these things and how we usually instantiate event loops.
The upshot of this is that global event loops will cause a crash before
any undefined behavior is hit.
The event loop is responsible for handling POSIX signals while it's
running. The signal handler adds the signals to a wake pipe which is
then read after the select'ing code in wait_for_event. Problems happen,
however, when another signal comes in after the select wake: the signal
will interrupt the next syscall, the `read` from the wake pipe, and the
resulting EINTR in wait_for_event causes the program to crash. This is
undesirable. Instead, we want to retry reading as long as we're
interrupted.
After the previous change, the wake pipe was only being created on the
main thread by the main event loop. This change utilizes a flag to
always initialize the wake pipe on other threads. Because the pipe is
quite expensive (it will count towards the file descriptor limit, for
instance), we do the initialization "lazily": Only when an event loop is
constructed and it notices that there hasn't been a wake pipe created on
its thread, it will create the pipe. Conversely, this means that there
are no pipes on threads that never use an event loop.
The event loop system was previously very singletony to the point that
there's only a single event loop stack per process and only one event
loop (the topmost) can run at a time. This commit simply makes the event
loop stack and related structures thread-local so that each thread has
an isolated event loop system.
Some things are kept at a global level and synchronized with the new
MutexProtected: The main event loop needs to still be obtainable from
anywhere, as it closes down the application when it exits. The ID
allocator is global as IDs should not be shared even between threads.
And for the inspector server connection, the same as for the main loop
holds.
Note that currently, the wake pipe is only created by the main thread,
so notifications don't work on other threads.
This removes the temporary mutex fix for notifiers, introduced in
0631d3fed5 .
MutexProtected mirrors the identically-named Kernel primitive and can be
used to synchronize access to any object that might not be thread safe
on its own. Synchronization is done with a simple mutex, so access to a
MutexProtected object is potentially blocking.
Mutex now has an internal nesting variable which is there to harden it
against lock-unlock ordering issues (e.g. double unlocking).
This changes LibTimeZone to read the current time zone from the file
/etc/timezone rather than using tzset/tzname. Instead, in a subsequent
commit, LibC's time methods will be changed to used LibTimeZone to
retrieve the system time zone.
Also add an API to set the system time zone. This method is only allowed
when running within Serenity.
This is no longer needed now that LibTimeZone is included within LibC.
Remove the direct linkage so that others do not mistakenly copy-paste
the CMakeLists text elsewhere.
LibTimeZone will be needed directly within LibC for functions such as
localtime(). This change adds LibTimeZone directly within LibC, so that
LibTimeZone isn't its own .so library anymore.
LibTimeZone itself is compiled as an object library to make it easier to
give it generator-specific compilation flags.
This patch adds support for drawing triangular waves.
For now those can only be horizontal, but as they are intended for
underlining text, it's an okay way to handle this.
We first create the /dev/devctl based on the information from the SysFS.
Then, we create block devices and character devices based on the events
we read from that device.
If we run out of horizontal space when placing floating objects,
we now perform a "break" and continue with a new line of floats
below the bottommost margin edge of the current line.
This is definitely not 100% to-spec yet, but a huge improvement
on ACID1 already. :^)
This patch adds a BFC::FloatSideData struct so we can contain left and
right floating object layout state in a struct. This is preparation for
adding more per-side state.
This is poorly factored. TextNode needs to know whether the most
recently inserted fragment on the current line was empty or ended in
whitespace. This is used when deciding what to do with leading
whitespace in text nodes.
Let's keep this working for now, but we should eventually sort this out
and make text chunk iteration not depend on this information.
When collapsing whitespace, we can skip over all-whitespace chunks at
the start of each line, and immediately following fragments that
themselves end in whitespace.
Now that we build lines incrementally, we no longer need the atomic line
splitting API.
The new InlineLevelIterator and LineBuilder setup does have some
regressions from the old behavior, but we can deal with them as we go.
We don't want to descend into inline-block containers when iterating
inline-level items, since that would make the inline-level children of
the inline-block bubble up to the containing block of the inline-block.
This resolves a long-standing architectural problem in LibWeb that made
it unable to place CSS floating objects correctly due to not having
final vertical position information when computing the amount of
available horizontal space for each line.
This patch adds a new mechanism that allows InlineFormattingContext to
build line boxes incrementally instead of all-in-one go.
Incremental build will eventually allow much better support for CSS
floating objects.
`auto count = 0;` will declare `count` as a `signed int`.
We don't want that since `count` is used to count the occurence of an
element in an `AK::Vector` that can have up to `SIZE_MAX` elements;
`SIZE_MAX` can overflow a `signed int` more than 4 billion times.
Problem:
- The statistics overlay period is hardcoded to 500 ms. This time is
very short and can result in the values being very "jumpy".
Solution:
- Increasing this value can result in more steady values which is
useful when trying to evaluate the performance impact of a change. A
new config value is offered in `Config.h` to let the developer
change to any value desired.
This moves LibWeb to using the list of hidden elements from the spec.
Concretely, the following things are now explicitly marked
`display: none` in addition to before:
- elements with the `hidden` attribute
- area
- base
- basefont
- datalist
- param
- rp
The spec also wants `noframes` and `noembed` to be `display: none`,
but since support for frames and embeds doesn't exist yet, these
are omitted for now.
With this, everyone's favorite website http://45.33.8.238/ no longer
displays spans with attribute hidden. (Whitespace handling still
looks a bit off though.)
vdbgln() was responsible for ~10% of samples on pv's flamegraph for
RequestServer (under request_did_finish) when loading github.com in
Browser and recording a whole-system profile. This makes that almost
completely disappear.
This has some risks as it can (attempt to) load arbitrary files on the
filesystem but for now it can only load local files which do go through
normal file operations. But let's enable it for now to see what we can
do with it.
Now that module loading is implemented this just works :^).
Since ShadowRealm explicitly passed a null ScriptOrModule we attempt to
get the top most ScriptOrModule in HostImportModuleDynamically.
This won't work in general as the web specifies other behavior but for
LibJS there must always be an active script to call
HostImportModuleDynamically.
This allows us to load modules from scripts.
This can be dangerous as it can load arbitrary files. Because of that it
fails and throws by default. Currently, only js and JavaScriptTestRunner
enable the default hook.
This also adds tests to test-js which test module code. Because we
form a spec perspective can't "enter" a module this is the easiest way
to run tests without having to modify test-js to have special cases for
modules.
To specify modules in test-js we use the extension '.mjs' this is to
ensure the files are not executed. We do still want to lint these files
so the prettier scripts have changed to look for '.mjs' files as well.
This loads modules with relative paths from the referencing module.
In this commit the only way to run a module is via the interpreter
which can link and evaluate a module (and all its dependencies).
The big changes are:
- Allow strings as Module{Export, Import}Name
- Properly track declarations in default export statements
However, the spec is a little strange in that it allows function and
class declarations without a name in default export statements.
This is quite hard to fully implement without rewriting more of the
parser so for now this behavior is emulated by faking things with
function and class expressions. See the comments in
parse_export_statement for details on the hacks and where it goes wrong.
This also refactors interpreter creation to follow
InitializeHostDefinedRealm, but I couldn't fit it in the title :^)
This allows us to follow the spec much more closely rather than being
completely ad-hoc with just the parse node instead of having all the
surrounding data such as the realm of the parse node.
The interpreter creation refactor creates the global execution context
once and doesn't take it off the stack. This allows LibWeb to take the
global execution context and manually handle it, following the HTML
spec. The HTML spec calls this the "realm execution context" of the
environment settings object.
It also allows us to specify the globalThis type, as it can be
different from the global object type. For example, on the web, Window
global objects use a WindowProxy global this value to enforce the same
origin policy on operations like [[GetOwnProperty]].
Finally, it allows us to directly call Program::execute in perform_eval
and perform_shadow_realm_eval as this moves
global_declaration_instantiation into Interpreter::run
(ScriptEvaluation) as per the spec.
Note that this doesn't evalulate Source Text Modules yet or refactor
the bytecode interpreter, that's work for future us :^)
This patch was originally build by Luke for the environment settings
object change but was also needed for modules. So I (davidot) have
modified it with the new completion changes and setup for that.
Co-authored-by: davidot <davidot@serenityos.org>
Instead of leaking all capture groups and selectively clearing some,
simply avoid leaking things and only "define" the ones that need to
exist.
This *actually* implements the capture groups ECMA262 quirk.
Also adds the test removed in the previous commit (to avoid messing up
test runs across bisects).
This partially reverts commit c11be92e23.
That commit fixes one thing and breaks many more, a next commit will
implement this quirk in a more sane way.
Previously, we weren't ever populating the read buffer in read(), which
was making the BufferedHelper useless, how silly :^). This introduces
a buffer refill when we have run out of buffered samples, restoring
FlacLoader performance from the new low of 200% (directly before this
commit) to the old level of ~1400%.
MemoryStream is the Core::Stream API's streamlike access to a chunk of
memory, mimicking AK::DuplexMemoryStream. The implementation is very
similar, except that no APIs except the SeekableStream operations
currently exist. This will be fine for the first users and can be
expanded upon later.
BigEndianInputBitStream is the Core::Stream API's bitwise input stream
for big endian input data. The functionality and bitwise read API is
almost unchanged from AK::BitStream, except that this bit stream only
supports big endian operations.
As the behavior for mixing big endian and little endian reads on
AK::BitStream is unknown (and untested), it was never done anyways. So
this was a good opportunity to split up big endian and little endian
reading.
Another API improvement from AK::BitStream is the ability to specify
the return type of the bit read function. Always needing to static_cast
the result of BitStream::read_bits_big_endian into the desired type is
adding a lot of avoidable noise to the users (primarily FlacLoader).
Previously, Browser loaded icons from the disk every time an icon
was set. In addition to making more calls to the disk and decoding
more images, this makes error propagation impossible. This change
moves all icon loading to the start of the program.
Previously, Button::set_icon required moving the bitmap into the
button, preventing the same bitmap from being used by multiple
buttons at once. While this works for buttons that are created once,
any button that is dynamically added would require the same bitmap to
be loaded every single time. In addition to being ineffecient, this
also makes error checking more difficult.
With this change, a bitmap can be loaded once, and passed to multiple
buttons.
Mostly slapping "timeZone: UTC" on DateTimeFormat tests (we have other
tests for specific time zones). Also pick dates that are not on DST
boundaries in some time zones where that matters.
This hides the method Group::add_group() on both MacOS and OpenBSD since
the function putgrent(), which is essential for add_group() to work, is
not available on these OSes.
This was easily done, as the Kernel and Userland don't actually share
any of the APIs exposed by it, so instead the Kernel APIs were moved to
the Kernel, and the Userland APIs stayed in LibKeyboard.
This has multiple advantages:
* The non OOM-fallible String is not longer used for storing the
character map name in the Kernel
* The kernel no longer has to link to the userland LibKeyboard code
* A lot of #ifdef KERNEL cruft can be removed from LibKeyboard
Previously we were jumping to the new end of the previous block (created
by the newly inserted ForkStay), correct the offset to jump to the
correct block as shown in the comments.
Fixes#12033.
This mirrors the previous default in Core::LocalSocket, and is the safer
default anyway. This prevents fds from living on in other processes when
exec() is called in certain programs such as Assistant.
Fixes#12029.
The constant value for GL_DECAL is 0x2101 instead of 0x2102.
This was tripping up Half-Life when making the water texture
transparent when under water. The Half-Life port uses its own OpenGL
header, meaning this error wasn't hidden by us.
Previously, Emulator::virt$execve would not report ENOENT and EACCES
when the binary to be executed was nonexistent or not executable. This
broke the execp family of functions, which rely on ENOENT being reported
in order to know that they should continue searching $PATH.
Emulator::virt$execve would construct command lines such as
`/bin/UserspaceEmulator echo -- hello` instead of
`/bin/UserspaceEmulator -- echo hello`, which naturally caused problems.
This commit moves the "--" to the correct place.
... into QuickLaunchEntry class. It will be used to implement adding
plain executables to the taskbar. For now, it adds TRY() error handling
to app launching :^)
test-js crashes with a segmentation fault when running on macOS on Arm.
Increasing the margin in the test in did_reach_stack_space_limit() to
32 * KiB makes the tests pass. To simplify the code, this is applied
independently of platform, and the previous test for use of an address
sanitizer is removed.
In querySelector(All)'s use of "Match a Selector Against a Tree", it
passes in the node the function was called on as the "optional scoping
root", which causes it and the nodes which aren't descendants of it
to be excluded from the list of possible nodes to match against.
For us, this is the equivalent of using the non-inclusive variant of
`for_each_in_subtree_of_type`.
This was tripping up the node re-ordering logic of d3 and would cause
it to try and reinsert nodes into their parent, causing an exception
to be thrown.
Note that this should be shadow-including, but we don't currently have
shadow-including tree traversal as per https://dom.spec.whatwg.org/#concept-shadow-including-tree-orderhttps://drafts.csswg.org/selectors-4/#match-a-selector-against-a-treehttps://dom.spec.whatwg.org/#scope-match-a-selectors-string
This is wrong because we have already set the fd in the
PosixSocketHelper, and the destructor of the respective Socket class
will close the fd for us. With the manual closing of the fd, we attempt
to close the same fd twice which results in a crash.
Thanks to stelar7 for noticing this bug.
Since all users of the old API are now removed, this commit removes all
the methods that returned raw file descriptors, in favor of returning
`ErrorOr<NonnullRefPtr<Core::File>`.
This commit also removed the redundant `filename` parameter from
`GLContextWidget::load_file`, since the filename is already stored
within the file itself.
The current implementation is a bit of a hack since we also want to keep
around the previous variants for now, but will be cleaned up later once
all applications have been ported to the new API.
Previously when opening a crash report for HackStudio, the
`unveil("/bin/HackStudio", "rx")` call was failing because of the
earlier `unveil(executable_path.characters(), "r")` call requesting only
"r" permissions for it. This patch handles this specific case, so you
can crash HackStudio to your heart's content. :^)
Also, we were unveiling the executable path twice, once manually and
once implicitly as part of the coredump's libraries, so we now check for
the latter and avoid it.
Thanks to Daniel for noticing what was right in front of me and I didn't
see!
Co-authored-by: Daniel Bertalan <dani@danielbertalan.dev>
Despite looking like it was still needed, it was only used for passing
to other calls to Length::resolved() recursively. This makes the
various `foo.resolved().resolved()` calls a lot less awkward.
(Though, still quite awkward.)
I think we'd need to separate calculated lengths out to properly tidy
these calls up, but one yak at a time. :^)
A lot of this is quite ugly, but it should only be so until I remove
Length::Type::Percentage entirely. (Which should happen later in this
PR, otherwise, yell at me!) For now, a lot of things have to be
resolved twice, first from a LengthPercentage to a Length, and then
from a Length to a pixel one.
The flexbox logic confuses me so regressions are possible, though our
test page looks the same as before so it should be fine.
Renamed FlexBasis::Length -> LengthPercentage too, for clarity.
This does undo the changes in 88c32836d8,
which accounted for our bitmap fonts being a different size than the
`font-size` property requests. I think this would be better handled
inside Length::to_px(), which would then apply to all font-size-relative
lengths (eg, em and rem) instead of only for the line-height property.
Layout::Node still treats border radii as having a single value instead
of horizontal and vertical, but one less hack is nice, and helps with
conversion to LengthPercentage. :^)
