They're mostly used in literal random data, so it isn't like
there is a high demand for it, but it's cool to have more complete
implementation anyway. :^)
This improves the look of tabs and their focus rects. In particular, the
concept of a "text rect" is removed, and whatever tab content area is
left over after the icon and close button are added is used as the area
to draw the text into. This approach is simpler than having a separate
text rect.
Shades are colors darker than the color, tints are colors lighter.
This is helpful in places where we need a bunch of similar colors
with some small differences.
The common thin-cap button look (1px highlight, 2px shadow) looks nice
on regular buttons, but the scrollbar didn't feel quite right.
This patch adds 1px of offset to the highlight, giving it a thick-cap
look (which I have named Gfx::ButtonStyle::ThickCap) :^)
Once canonical extensions are implemented, the number of:
if (optional_string.has_value() {
builder.append('-');
builder.append(optional_string->to_lowercase_string());
}
Will be quite large. This commit just adds a helper lambda to handle
this pattern to prevent this function from becoming even more enormous.
This is preparatory work to read locale extensions. The parser currently
enforces that the entire string is consumed. But to parse extensions,
parse_unicode_locale_id() will need parse_unicode_language_id() to just
stop parsing on the first segment that does not match the language ID
grammar. It will also need to know where the parsing stopped. Both of
these needs are fulfilled by GenericLexer.
The caveat is that we can no longer simply split the parsed string on
separator characters. So parse_unicode_language_id() now operates as a
small state machine.
This is needed so all headers and files exist on disk, so that
the sonar cloud analyzer can find them when executing the compilation
commands contained in compile_commands.json, without actually building.
Co-authored-by: Andrew Kaster <akaster@serenityos.org>
This commit adds a new process method to all Decoder subclasses which
do what to_utf8 used to do, and allows callers to customize the handling
of individiual UTF-8 code points through a callback. Decoder::to_utf8
now uses this API to generate a string via StringBuilder, preserving the
original behavior.
This always subtracted the glyph width of a space, despite isspace
also accepting newlines and a few other characters. It now also uses
AK/CharacterTypes.h. :^)
Non-printable characters should always have a width of 0. This is not
true for some characters like tab, but those can be exempted as the need
arises. Doing this here saves us from a bunch of checks in any place
that needs to figure out glyph widths for text which can contain
non-printable characters.
This more clearly expresses the purpose of this flag. Since only
CSS::WhiteSpace::Nowrap sets this value to false and it does not respect
linebreaks, this made the most sense as a flag name.
This commit refactors the text chunking algorithm used in
TextNode::ChunkIterator. The m_start_of_chunk member parameter has been
replaced with a local variable that's anchored to the current iterator
at the start of every next() call, and the algorithm is made a little
more clear by explicitly separating what can and cannot peek into the
next character during iteration.
We don't need transitions for either of these:
- Adding the 'name' property to a constructor object
- Adding the 'constructor' property to its prototype object
- Replace the misleading abuse of the m_transitions_enabled flag for the
fast path without lookup with a new m_initialized boolean that's set
either by Heap::allocate() after calling the Object's initialize(), or
by the GlobalObject in its special initialize_global_object(). This
makes it work regardless of the shape's uniqueness.
- When we're adding a new property past the initialization phase,
there's no need to do a second metadata lookup to retrieve the storage
value offset - it's known to always be the shape's property count
minus one. Also, instead of doing manual storage resizing and
assignment via indexing, just use Vector::append().
- When we didn't add a new property but are overwriting an existing one,
the property count and therefore storage value offset doesn't change,
so we don't have to retrieve it either.
As a result, Object::set_shape() is now solely responsible for updating
the m_shape pointer and is not resizing storage anymore, so I moved it
into the header.
We don't need to be allocating Strings for these names during static
initialization. The C-string literals will be stored in the .rodata ELF
section, so they're not going anywhere. We can just wrap the .rodata
storage for the class names in StringViews and use those in Object
registration and lookup APIs.
Optimizations:
- Make sure `DT_SYMTAB` is a string view literal, instead of string.
- DynamicObject::HashSection::lookup_sysv_symbol should be using
raw_name() from symbol comparison to avoid needlessly calling
`strlen`, when the StrinView::operator= walks the cstring without
calling `strlen` first.
- DynamicObject::HashSection::lookup_gnu_symbol shouldn't create a
symbol unless we know the hashes match first.
In order to test these changes I enabled Undefined behavior sanitizer
which creates a huge amount of relocations, and then ran the browser
with the help argument 100 times. The browser is a fairly big app with
a few different libraries being loaded, so it seemed liked a good
target.
Command: `time -n 100 br --help`
Before:
```
Timing report:
==============
Command: br --help
Average time: 3897.679931 ms
Excluding first: 3901.242431 ms
```
After:
```
Timing report:
==============
Command: br --help
Average time: 3612.860107 ms
Excluding first: 3613.54541 ms
```
This allows us to remove all the add_subdirectory calls from the top
level CMakeLists.txt that referred to targets linking LagomCore.
Segregating the host tools and Serenity targets helps us get to a place
where the main Serenity build can simply use a CMake toolchain file
rather than swapping all the compiler/sysroot variables after building
host libraries and tools.
Gather the custom commands for each of the 6 bindings generated targets
for libjs_js_wrapper invocations into some lists so that we can foreach
over the lists instead of having 6 copy pasted commands with one or two
things modified for each one.
Additional refactoring, use target_sources command to inform CMake about
additional source files for LibWeb, but only after it's been declared as
a library via add_library. Also avoid use of the write_if_different
script and use cmake -E copy_if_different instead. This lets us express
the actions in rules that CMake understands without going to an external
source file. It exposes a few optimization opportunities for the code
generators to accept an output filename instead of always going to
stdout.
Moving this helper CMake file to the centralized Meta/CMake folder helps
to get a better grasp on what extra files are required for the build,
and what files are generated.
While we're at it, don't use add_compile_definitions for
ENABLE_UNICODE_DATA, which only needs to be seen by LibUnicode sources.
In general, I think `opt == x` looks much nicer than
`opt.has_value() && opt.value() == x`, so I'm updating
the remaining few instances I could find with some regex
magic in my search.
All audio applications (aplay, Piano, Sound Player) respect the ability
of the system to have theoretically any sample rate. Therefore, they
resample their own audio into the system sample rate.
LibAudio previously had its loaders resample their own audio, even
though they expose their sample rate. This is now changed. The loaders
output audio data in their file's sample rate, which the user has to
query and resample appropriately. Resampling code from Buffer, WavLoader
and FlacLoader is removed.
Note that these applications only check the sample rate at startup,
which is reasonable (the user has to restart applications when changing
the sample rate). Fully dynamic adaptation could both lead to errors and
will require another IPC interface. This seems to be enough for now.
Two new ioctl requests are used to get and set the sample rate of the
sound card. The SB16 device keeps track of the sample rate separately,
because I don't want to figure out how to read the sample rate from the
device; it's easier that way.
The soundcard write doesn't set the sample rate to 44100 Hz every time
anymore, as we want to change it externally.
This AO is required for a bunch of PlainTime related methods.
As part of this change the `TemporalTime` record was renamed to
`UnregulatedTemporalTime` and a new `TemporalTime` record that matches
the other Temporal parse result records was added. This also has the
added benefit of getting rid of a would be round-trip cast from integer
to double and back in `ParseTemporalTimeString`.
This commit is preemptive to upcoming commits which add more subtags to
the CLDR generator. Rather than generating a giant HashMap containing
all data, generate more (smaller) Array-based tables. This mimics the
UCD generator. This also allows simpler lookups at runtime since we can
generate index-based lookups into the smaller tables rather easily.
Without this change, adding the remaining locale subtags would result
in the generation and compilation of UnicodeLocale.cpp taking about 30s
on my machine. With this change, it takes about half that. Additionally,
the size of the generated file reduces by about 1.5MB.
The UCD set of data contained a very small subset of all locales just to
handle some special casing rules. This enumeration will be needed within
the CLDR generator as well. So rather than duplicate the enum, remove it
from the UCD generator in favor of the full list of locales known by the
CLDR generator.
This patch adds a Config::Listener abstract class that anyone can
inherit from and receive notifications when configuration values change.
We don't yet monitor file system changes, so these only work for changes
made by ConfigServer itself.
In order to receive these notifications, clients must monitor the domain
by calling monitor_domain(). Only pledged domains can be monitored.
Note that the client initiating the change does not get notified.
Note that only option type=region is really implemented. Other types
will resort to the fallback option. This prototype method will be able
to implement other type options once LibUnicode supports more.
There is notably FIXME notations in this commit regarding Unicode locale
extensions. We are not parsing extensions (or private use extensions) at
all yet.
ECMA-402 requires validating user input against the EBNF grammar for
Unicode locales described in TR-35: https://www.unicode.org/reports/tr35
This commit adds validators for that grammar, as well as other helper to
e.g. canonicalize a locale string.
The Unicode standard publishes a database known as the Common Locale
Data Repository (CLDR). This is a massive set of data from which anyone
implementing Unicode's Technical Standard #35 may generate their
implementation: https://www.unicode.org/reports/tr35/
This commit updates LibUnicode to download the compressed database and
extract a small subset. That subset is used to generate a list of
available locales and the territories (AKA regions) associated with each
locale.
This controls how fetched texels are combined with the color that was
produced by a preceding texture unit or with the vertex color if it is
the first texture unit.
Currently only a small subset of possible combine modes is implemented
as required by glquake.
This sets the length of a row for the image to be transferred. This
value is measured in pixels. When a rectangle with a width less than
this value is transferred the remaining pixels of this row are skipped.
Previously it was returning an "auto" length. This caused all the new
"initial" values to effectively turn into auto values long before layout
had a chance to resolve them.
This broke replaced elements with intrinsic size but no specified width
or height, and is the reason that Mr. ACID2 temporarily lost his eyes.
This API lets applications specify which configuration domains they
will be accessing throughout their lifetime. It works similarly in
spirit to the kernel's pledge().
You cannot pledge_domains() more than once, and once you have used it,
it's no longer possible to access any other configuration domain.
This is obviously just a first cut of this mechanism, and we may need
to tweak it further as we go.
It's not possible to connect to ConfigServer without having an event
loop available. This VERIFY makes it much easier to understand why
things are not working. :^)
This static bool getter can be used to VERIFY that an event loop exists,
in situations where one is expected.
This is helpful if the absence of an event loop would generate strange
and/or loud errors that don't immediately point to this as a cause.
ConfigServer is an IPC service that provides access to application
configuration and settings. The idea is to replace all uses of
Core::ConfigFile with IPC requests to ConfigServer.
This first cut of the API is pretty similar to Core::ConfigFile.
The old:
auto config = Core::ConfigFile::open_for_app("App");
auto value = config->read_entry("Group", "Key");
The new:
auto value = Config::read_string("App", "Group", "Key");
ConfigServer uses the ~/.config directory as its backing store
and all the files remain human-editable. :^)
Calling sigprocmask() through the PLT requires setting the ebx register
to the address of the global offset table, otherwise chaos ensues. Also
the value of the ecx register was assumed to be preserved across that
function call despite the fact that it is caller-saved in the x86
calling convention.
This isn't 100% spec complaint, as it should use glyph_height()
depending on what the value of the writing-mode is, but we haven't
implemented it yet, so I think it'll be good enough for now.
This can be tested in https://wpt.live/css/css-values/ch-unit-008.html
Other css-unit tests fail as:
- 001 shows an issue related to a renderer (looks to me like you can't
pass a width and height property to a span -- adding `display: block`
to it passes the test),
- 002-004 and 009-012 use mentioned writing-mode,
- 016-017 loads custom fonts, which we also don't support (yet).
When property() previously would have returned an InitialStyleValue, we
now look up what the initial value would be, and return that instead.
We also intercep 'inherit', but inheritance is not implemented yet so we
just return nothing.
This does cause a regression on Acid2: The eyes no longer appear, and I
am not sure why. :^(
Since we have initial-value data in Properties.json already, it makes
sense to use that instead of needing to duplicate the same information
in ComputedValues.h
However, converting a StyleValue to the kind of types used in
InitialValues is non-trivial. So this may or may not actually be useful.
This bug was discovered via OSS fuzz, it's possible to fall through
to this assert with a char_size == 1, so we need to account for that
in the VERIFY(..).
Repro test case can be found in the OSS fuzz bug:
https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=37296
This patch adds OutOfProcessWebView::run_javascript(StringView).
This can be used by the OOPWV embedder to execute arbitrary JavaScript
in the top-level browsing context on the WebContent process side.
Added a test to ensure the behavior stays the same.
We now throw on a direct usage of an escaped keywords with a specific
error to make it more clear to the user.
This iterates the fragments of the containing block, and paints their
outlines if they are descendants of the InlineNode.
If multiple fragments are adjacent, eg:
```html
<span><b>Well</b> hello <i>friends!</i></span>
```
...then we get a double-thick outline between "Well", " hello " and
"friends!", but we can come back to this after we implement
non-rectangular outlines for the `outline` CSS property.
These should all have a name with an empty string. Not only does test262
verify this, but it also verifies that (for the executor) the name
property is defined after the length property.
PerformPromiseAll, PerformPromiseAny, PerformPromiseAllSettled, etc, all
have very similar iteration loops. To avoid duplicating this rather
large block of code, extract the common functionality into a separate
method.
The element-resolving functions on the Promise constructor are all very
similar. To prepare for more of these functions to be implemented, break
out common parts into a base class.
Ctrl+Shift+Left would add the word before the cursor to the selection,
but for some reason Ctrl+Shift+Right didn't add the word after the
cursor to the selection.
1. Move htonl() etc. from <arpa/inet.h> to <netinet/in.h> (which
<arpa/inet.h> includes).
The htonl(), htons(), ntohl(), and ntohs() functions shall be
available as described in <arpa/inet.h>. Inclusion of the
<netinet/in.h> header may also make visible all symbols from
<arpa/inet.h>.
- POSIX
2. Define IN6_IS_ADDR_LOOPBACK() and IN6_IS_ADDR_V4MAPPED()
For some reason X/OPEN requires that fd_set has a field fds_bits. Xproto
requires either fds_bits or _fds_bits to be present, so the field 'bits'
was renamed 'fds_bits'
This patch adds support for opening a ConfigFile using a file
descriptor rather than trying to open a the file by name directly.
In contrast to the previous implementation, ConfigFile now always keeps
a reference to an open File and does not reopen it for writing.
This requires providing an additional argument to open functions if a
file gets opened based on its name and the user of the api intends to
write to the file in the future.
Previously, the preprocessor first split the source into lines, and then
processed and lexed each line separately.
This patch makes the preprocessor first lex the source, and then do the
processing on the tokenized representation.
This generally simplifies the code, and also fixes an issue we
previously had with multiline comments (we did not recognize them
correctly when processing each line separately).
Classes reading and writing to the data heap would communicate directly
with the Heap object, and transfer ByteBuffers back and forth with it.
This makes things like caching and locking hard. Therefore all data
persistence activity will be funneled through a Serializer object which
in turn submits it to the Heap.
Introducing this unfortunately resulted in a huge amount of churn, in
which a number of smaller refactorings got caught up as well.
This patch provides very basic, bare bones implementations of the
INSERT and SELECT statements. They are *very* limited:
- The only variant of the INSERT statement that currently works is
SELECT INTO schema.table (column1, column2, ....) VALUES
(value11, value21, ...), (value12, value22, ...), ...
where the values are literals.
- The SELECT statement is even more limited, and is only provided to
allow verification of the INSERT statement. The only form implemented
is: SELECT * FROM schema.table
These statements required a bit of change in the Statement::execute
API. Originally execute only received a Database object as parameter.
This is not enough; we now pass an ExecutionContext object which
contains the Database, the current result set, and the last Tuple read
from the database. This object will undoubtedly evolve over time.
This API change dragged SQLServer::SQLStatement into the patch.
Another API addition is Expression::evaluate. This method is,
unsurprisingly, used to evaluate expressions, like the values in the
INSERT statement.
Finally, a new test file is added: TestSqlStatementExecution, which
tests the currently implemented statements. As the number and flavour of
implemented statements grows, this test file will probably have to be
restructured.
The implemtation of the Value class was based on lambda member variables
implementing type-dependent behaviour. This was done to ensure that
Values can be used as stack-only objects; the simplest alternative,
virtual methods, forces them onto the heap. The problem with the the
lambda approach is that it bloats the Values (which are supposed to be
lightweight objects) quite considerably, because every object contains
more than a dozen function pointers.
The solution to address both problems (we want Values to be able to live
on the stack and be as lightweight as possible) chosen here is to
encapsulate type-dependent behaviour and state in an implementation
class, and let the Value be an AK::Variant of those implementation
classes. All methods of Value are now basically straight delegates to
the implementation object using the Variant::visit method.
One issue complicating matters is the addition of two aggregate types,
Tuple and Array, which each contain a Vector of Values. At this point
Tuples and Arrays (and potential future aggregate types) can't contain
these aggregate types. This is limiting and needs to be addressed.
Another area that needs attention is the nomenclature of things; it's
a bit of a tangle of 'ValueBlahBlah' and 'ImplBlahBlah'. It makes sense
right now I think but admit we probably can do better.
Other things included here:
- Added the Boolean and Null types (and Tuple and Array, see above).
- to_string now always succeeds and returns a String instead of an
Optional. This had some impact on other sources.
- Added a lot of tests.
- Started moving the serialization mechanism more towards where I want
it to be, i.e. a 'DataSerializer' object which just takes
serialization and deserialization requests and knows for example how
to store long strings out-of-line.
One last remark: There is obviously a naming clash between the Tuple
class and the Tuple Value type. This is intentional; I plan to make the
Tuple class a subclass of Value (and hence Key and Row as well).
This is an interesting quirk that occurs due to us using the x87 FPU
when Serenity is compiled for the i386 target. When we calculate our
depth value to be stored in the buffer, it is an 80-bit x87
floating point number, however, when stored into the DepthBuffer,
this is truncated to 32 bits. This 38 bit loss of precision means
that when x87 `FCOMP` is eventually used here the comparison fails.
This could be solved by using a `long double` for the depth buffer,
however this would take up significantly more space and is completely
overkill for a depth buffer. As such, comparing the first 32-bits of
this depth value is "good enough" that if we get a hit on it being
equal, we can pretty much guarantee that it's actually equal.
For example, consider the following pattern:
new RegExp('\ud834\udf06', 'u')
With this pattern, the regex parser should insert the UTF-8 encoded
bytes 0xf0, 0x9d, 0x8c, and 0x86. However, because these characters are
currently treated as normal char types, they have a negative value since
they are all > 0x7f. Then, due to sign extension, when these characters
are cast to u64, the sign bit is preserved. The result is that these
bytes are inserted as 0xfffffffffffffff0, 0xffffffffffffff9d, etc.
Fortunately, there are only a few places where we insert bytecode with
the raw characters. In these places, be sure to treat the bytes as u8
before they are cast to u64.
RegExp.prototype.compile will require invoking RegExpInitialize on an
already-existing RegExpObject. Break up RegExpCreate into RegExpAlloc
and RegExpInitialize to support this.
Currently just sets the renderer option for what polygon mode we
want the rasterizer to draw in. GLQuake only uses `GL_FRONT_AND_BACK`
with `GL_FILL` )which implies both back and front facing triangles
are to be filled completely by the rasterizer), so keeping this as
a small stub is perfectly fine for now.
With the new parser, we started interpreting the `opacity` property as a
string value, which made it turn into `auto` and so anything with
opacity ended up not visible (e.g the header on google.com)
This patch restores our old behavior for `opacity` by interpreting it
as a numeric value with optional decimals.
Instead of loading every icon, only load the filetype image icon if it
hasn't been already. This icon is used by IconViews that need to lazily
load thumbnails, which don't need any of the other icon types.
Spending the time to load the unneeded images was causing delays to
first paint in BackgroundSettings.
For example, "property.br\u{64}wn" should resolve to "property.brown".
To support this behavior, this commit changes the Token class to hold
both the evaluated identifier name and a view into the original source
for the unevaluated name. There are some contexts in which identifiers
are not allowed to contain Unicode escape sequences; for example, export
statements of the form "export {} from foo.js" forbid escapes in the
identifier "from".
The test file is added to .prettierignore because prettier will replace
all escaped Unicode sequences with their unescaped value.
Unfortunately, this requires a slight divergence in the way the capture
group names are stored. Previously, the generated byte code would simply
store a view into the regex pattern string, so no string copying was
required.
Now, the escape sequences are decoded into a new string, and a vector
of all parsed capture group names are stored in a vector in the parser
result structure. The byte code then stores a view into the
corresponding string in that vector.
This will allow regex::Lexer users to invoke GenericLexer consumption
methods, such as GenericLexer::consume_escaped_codepoint().
This also allows for de-duplicating common methods between the lexers.
This is primarily to be able to remove the GenericLexer include out of
Format.h as well. A subsequent commit will add AK::Result to
GenericLexer, which will cause naming conflicts with other structures
named Result. This can be avoided (for now) by preventing nearly every
file in the system from implicitly including GenericLexer.
Other changes in this commit are to add the GenericLexer include to
files where it is missing.
It can sometimes be difficult to tell from the debug.log and test stdout
which test was the last to run before the test runner hangs or exits the
QEMU instance unexpectedly.
Print out a start message before each test is executed, along with a
progress message indicating which test out of how many tests we're about
to run.
This brings `glGetFloatv` more inline with the other `glGet`
functions. We should prevent crashing in the driver as much as
possible and instead let the application deal with the generated
GL error.
Instead of constructing a String and converting that to a PropertyName
on the fly, we can just leverage CommonPropertyNames, add a couple more
and directly pass ready-to-use PropertyNames with pre-allocated Strings.