This commit upstreams most of the C++ bits of the LibJS test262 runner
at https://github.com/linusg/libjs-test262/, specifically everything but
the main.cpp file serving as the actual executable.
Since all of these are just regular JS objects, I opted to put them in
LibJS itself, in a new Contrib/ directory like many other projects have
one. Other code that can end up there in the future is the runtime for
esvu, which might even share some functionality with test262's $262
object.
The code has been copied verbatim, and only a small number of changes
have been made:
- Putting everything into the JS::Test262 namespace
- Removing now redundant JS namespace prefixes
- Updating includes to use absolute <LibJS/...> paths
- Updating the SPDX-License-Identifier comments from MIT to BSD-2-Clause
I gained permission to change the license and upstream these changes
from all the major contributors to this code: Ali, Andrew, David, Idan.
The removal of the code from the source repository is here:
https://github.com/linusg/libjs-test262/pull/54
This is only the first step, the goal is to eventually upstream the
actual libjs-test262-runner executable and supporting Python scripts
into SerenityOS as well.
Unlike BFC root blocks with height:auto, when the block *isn't* a BFC
root, we don't have to look for the "bottommost" block-level child and
determine the width from that.
Instead, we should just look at the last in-flow block-level child.
This was already indicated in the spec comment next to the code, but
the code itself was wrong.
This makes the body element on Acid3 have the correct height. It also
introduces a small regression on Acid2 that we'll have to track down.
Preserve floating point precision and delay rounding until the last
moment when figuring out where to paint background layers. This fixes an
issue on Acid3 where a thin sliver of red was visible because the
background X position was incorrectly rounded by 1px.
The find widget appears on Ctrl+F.
It uses the GUI::TextEditor search API to search for text, which also
takes care of highlighting the search results.
This adds a search API to TextEditor.
The API that is similar to "find_text" of TextDocument (which is used
internally to do the search).
All search results (as well as the current one) are highlighted with
a "span collection", which is pretty neat :^)
TextDocument::set_spans() now also takes a "span collection index"
argument.
TextDocument keeps a map between a span collection index and its spans.
It merges the spans from all collections into a single set of spans
whenever set_spans() is called.
This allows us to style a document with multiple layers of spans, where
as previously we only supported a single layer of spans that was set
from the SyntaxHighlighter.
When parsing the "style" attribute on elements, we'd previously ask the
CSS parser for a PropertyOwningCSSStyleDeclaration. Then we'd create a
new ElementCSSInlineStyleDeclaration and transfer the properties from
the first object to the second object.
This patch teaches the parser to make ElementCSSInlineStyleDeclaration
objects directly.
By using enclosing_int_rect(), borders and backgrounds of boxes were
sometimes 1 pixel off, making things slightly larger than they should
be. Fix this by using to_rounded() instead of enclosing_int_rect().
There's definitely more of these type of issues lurking in the code,
and we'll get to them in time.
This feature needs a bit more work, so let's disable it by default.
Note that the shell will still use _complete_foo if it is defined
regardless of this setting.
POSIX describes WCHAR_MIN and WCHAR_MAX in stdint.h(0P), while
wchar.h(0P) only says "as described in stdint.h".
As there isn't a trivial path of "may make visible", just move it to a
shared header and include it from both files.
This fixes 2 bugs in our current implementation:
* Properties deleted during iteration were still being iterated
* Properties with the same name in both the object and it's prototype
were iterated twice
Previously, we only allowed floats to take up its own border box's worth
of horizontal space when laid out inside an IFC.
We should instead consume the full margin box horizonally. This fixes an
issue where a floated box on Acid3 had {width:20px; margin-right:-20px;}
but still consumed 20px of the previously available space, despite being
moved out of the way by its own negative margin.
When doing max-content layout, we were not committing newlines even
though we were supposed to due to white-space:pre*.
This broke the WPT harness due to a VERIFY() in ChunkIterator where we
were assuming the commit would always succeed.
Thanks to Orphis for reporting this! :^)
When the spec tells us to measure from the top content edge of a block,
that just means we should measure from Y=0. We don't need to go looking
for a child box with a negative top offset and measure from there.
Security handlers manage encryption and decription of PDF files. The
standard security handler uses RC4/MD5 to perform its crypto (AES as
well, but that is not yet implemented).
This was a small optimization to allow a stream object to simply hold
a reference to the bytes in a PDF document rather than duplicating
them. However, as we move into features such as encryption, this
optimization does more harm than good. This can be revisited in the
future if necessary.
When encountering a @font-face rule, StyleComputer will now fire off
a resource request and download the first source URL specified.
Once downloaded, we try to parse it as a TrueType font file, and if it
works, it's added to a cache in StyleComputer. This effectively makes
fonts per-document since every document has its own StyleComputer.
This is very unoptimized and could definitely use some caching, etc.
But it does work on Acid3. :^)
Previously we were passing raw UTF-8 bytes as code points, which caused
CSS content properties to display incorrect characters.
This makes bullet separators in Wikipedia templates display correctly.
This gets us a bit closer to the recommended algorithms in CSS 2.2 and
CSS Table Module 3.
A couple of table heavy websites (e.g. news.ycombinator.com,
html5test.com, etc.) now look quite okay. :^)
Currently this can parse XML and resolve external resources/references,
and read a DTD (but not apply or verify its rules).
That's good enough for _most_ XHTML documents as the HTML 5 spec
enforces its own rules about document well-formedness, and does not make
use of XML DTDs (aside from a list of predefined entities).
An accompanying `xml` utility is provided that can read and dump XML
documents, and can also run the XML conformance test suite.
When completing `ls -l` to add another short option, the invariant
length should be zero as we are not replacing anything with our
suggestion.
Also skip the initial dash if there already is one.
Fixes#13301.
Forgot to remove that in c0c9825f67, as
this function was no longer declared and used... Until the previous
commit.
This meant that pressing the F2 key after opening a file no longer
matched the current tab.
Besides from reusing more parts from the code, this allows us to call
the action using the F2 key. That is also the reason why we have
to reassign `m_tab_context_menu_sheet_view` on tab change.
On x86-64, `int64_t` is defined to be `long` (not `long long`) , so for
printing, the "l" format specifier has to be used instead of i686's
"ll".
A couple of these macros weren't updated when the x86-64 target was
added, so using them produced warnings like this:
> warning: format specifies type 'long long' but the argument has type
> 'int64_t' (aka 'long') [-Wformat]
>
> "DW_CFA_GNU_negative_offset_extended(%" PRId64 ")\n", offset);
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~
This commit changes the macros to be correct for both architectures, and
reorders them to be consistent and adds a couple missing ones for the
sake of completeness.
`delete` has to operate directly on Reference Records, so this
introduces a new set of operations called DeleteByValue, DeleteVariable
and DeleteById. They operate similarly to their Get counterparts,
except they end in creating a (temporary) Reference and calling delete_
on it.
When calling emit_load_from_reference with a MemberExpression, it is
only necessary to store the result of evaluating MemberExpression's
object when performing computed property lookup.
This allows us to skip unnecessary stores for identifier lookup.
For example, this would generate 3 unnecessary stores:
```
> Temporal.PlainDateTime.prototype.add
JS::Bytecode::Executable (REPL)
1:
[ 0] GetVariable 0 (Temporal)
[ 28] Store $2
[ 30] GetById 1 (PlainDateTime)
[ 40] Store $3
[ 48] GetById 2 (prototype)
[ 58] Store $4
[ 60] GetById 3 (add)
```
With this, it generates:
```
> Temporal.PlainDateTime.prototype.add
JS::Bytecode::Executable (REPL)
1:
[ 0] GetVariable 0 (Temporal)
[ 28] GetById 1 (PlainDateTime)
[ 38] GetById 2 (prototype)
[ 48] GetById 3 (add)
```
The body of for/in/of can contain an unconditional block terminator
(e.g. return, throw), so we have to check for that before generating
the Jump to the loop update block.
NewArray now only contains two elements maximum in `m_elements` to
indicate the range of registers to create the array from.
However, `m_element_count` still contains how many registers are in the
range and the stringifier was not updated to account for this. Thus, if
the range contained more than 2 registers, it would do a read OOB on
`m_elements`.
This makes it now just print the first and second entries in
`m_elements` in the format of `[<reg>-<reg>]`.
If we break out of the loop before we attempt to allocate again,
then we double free the memory pointed to by `name_path`.
Found by Static Analysis: Sonar Cloud
The names stdout / stderr are bound to conflict with existing
declarations when compiling against other LibC's. The build on OpenBSD
is broken for this reason at the moment.
Lets rename the members to more generic names to resolve the situation.
Previously, the whitespace collapsing code had a parameter telling it
whether the previous text node ended in whitespace. This was not
actually necessary, so let's get rid of it.
Previously LibLine accepted read callbacks while it was in the process
of reading input, this wasn't an issue as no async code was being
executed up until the Shell autocompletion came along.
Simply defer input processing while processing input to avoid causing
problems.
Fixes#13280.
We now support generating top-left submatrices from a `Gfx::Matrix`
and we move the normal transformation calculation into
`SoftGPU::Device`. No functional changes.
We were normalizing data read from vertex attribute pointers based on
their usage, but there is nothing written about this behavior in the
spec or in man pages.
When we implement `glVertexAttribPointer` however, the user can
optionally enable normalization per vertex attribute pointer. This
refactors the `VertexAttribPointer` to have a `normalize` field so we
can support that future implementation.
We were transforming the vertices' normals twice (bug 1) and
normalizing them after lighting (bug 2). In the lighting code, we were
then diverting from the spec to deal with the normal situation, which
is now no longer needed.
This fixes the lighting of Tux in Tux Racer.
This patch reimplements inset property resolution based on the new
CSS Positioned Layout specification. Nothing should change for
left/right insets, but we gain support for top/bottom. :^)
Relatively positioned boxes should not affect the *layout* of their
siblings. So instead of applying relative inset as a layout-time
translation on the box, we now perform the adjustment at the paintable
level instead.
This makes position:relative actually work as expected, and exposes some
new bugs we need to take care of for Acid2. :^)
Before this the flex layout didn't take into account the applied
borders or padding while laying out the items.
The child's top and left borders would get painted over the
parent's borders, also due to it not taking borders into account,
children with borders would overlap each other.
Due to it not taking padding into account, the children would get
drawn outside the parent element.
Some ISPs may MITM DNS requests coming from clients, changing the case
of domain name in response. LookupServer will refuse responses from
any DNS server in that case. This commit changes the behaviour to
perform a case-insensitive equality check.
If a C++ object already has a JS wrapper, we don't need to go through
the expensive type checks to figure out which kind of wrapper to create.
Instead, just return the wrapper we already have!
This gives a noticeable increase in smoothness on Acid3, where ~10% of
CPU time was previously spent doing RTTI type checks in wrap(). With
these changes, it's down to ~1%.
This reverts commit 2b2915656d.
While this adjustment is bogus, it is currently responsible for putting
CenterLeft aligned scalable text in the right position.
This is going to take a bunch of work to get right.
We've gotten ourselves into a bit of a mess by mixing pixel and point
sizes in multiple places. Step one towards getting out of this mess
is adding explicit accessors for the unit you're trying to fetch.
The core of the issue comes from bitmap fonts storing integer pixel
sizes and scaled (TTF) fonts storing float point sizes.
This is useful, for instance, in games in which you can switch held
items using the scroll wheel. In order to implement this, they
previously would have to either add a hard-coded division by 4, or look
up your mouse settings to adjust correctly.
This commit adds an MouseEvent.wheel_raw_delta_x() and
MouseEvent.wheel_raw_delta_y().
...but never allow the resulting height to become negative. This solves
an issue seen on Acid3 where elements with negative vertical margins
expanded the size of their height:auto container instead of shrinking
it, which is the correct behavior. This now works :^)
CSS 2.2 says "Horizontal margins never collapse."
So instead of collapsing them, we now add them together, which makes
negative margins between floating boxes work beautifully.
Instead of TextNode::ChunkIterator having two bool members to remember
things across calls to next(), this patch reorganizes the loop in next()
so that preserved newline/whitespace chunks are emitted right away
instead of in an awkward deferred way.
Instead of emitting a Text item with the "should_force_break" flag set
to true, newlines in newline-preserving text content now timply turn
into ForcedBreak items. This makes the <pre> element work again.
From the HTML spec:
Modulo platform conventions, it is suggested that the following
elements should be considered as focusable areas and be sequentially
focusable:
...
- button elements
- select elements
- textarea elements
...
Also add a spec link to the existing HTMLAnchorElement::is_focusable().
Note that this still doesn't allow triggering keyboard-focused buttons,
checkboxes, or radio buttons - we don't seem to run the expected
activation behavior for any of them.
From the HTML spec:
Modulo platform conventions, it is suggested that the following
elements should be considered as focusable areas and be sequentially
focusable:
...
- input elements whose type attribute are not in the Hidden state
...
Previously BGRx8888 was used, which produces artifacts with the new
antialiased window frames with border radii, which require alpha
blending whilst painting.
I believe this is all of them, but I may have missed some.
Several properties technically do not allow negative numbers but the
description says to accept these as valid, and then clamp them
afterwards to the desired range. As such, we don't reject them during
parsing.
This commit limits the autocomplete processes to effectively have
readonly access to the fs, and only enough pledges to get the dynamic
loader working.
This removes them from the main invocation example in --help, as well as
hides them from autocomplete results (we were previously special-casing
"help" and "version").
A program can either respond to `--complete -- some args to complete`
directly, or add a `_complete_<program name>` invokable (i.e. shell
function, or just a plain binary in PATH) that completes the given
command and lists the completions on stdout.
Should such a completion fail or yield no results, we'll fall back to
the previous completion algorithm.
This might not be entirely correct, but neither was using the completely
ad-hoc parse_html_length(), and this is the last user of that API so
let's move off of it.
For layers that require indirect painting (due to opacity, transform,
etc.) we create a nested PaintContext. Until now, that PaintContext
was created fresh without transferring all the state from the parent
PaintContext.
This commit moves the regular handling of links to the anchor elements'
activation behavior, and implements a few auxiliary algorithms as
defined by the HTML specification.
Note that certain things such as javascript links, fragments and opening
a new tab are still handled directly in EventHandler, but they have been
moved to handle_mouseup so that it behaves closer to how it would if it
was entirely up-to-spec.
Instead of using the absolute_rect(), use absolute_border_box_rect() -
at least for PaintableBox - and inflate it by 2px on each side.
This looks much nicer for text input elements, especially when they have
padding, which would be applied outside the focus rect previously.
It seems like this happens in quite some valid situations, so my
initially sensible failsafe doesn't make sense. As the buffer system is
hopefully gone soon, it won't be an issue in the future either way.
This was regressed at some point though I never saw it working.
Basically, while jump to slider works correctly it doesn't even get
actioned. While the user is clicking the slider it's very likely that a
buffer finishes playing and the callback for that changes the slider
value. This means that the user click just gets lost. There's some
additional weird behavior where values are lost in even more cases, so
an additional fix that is needed is to store the slider value in the
AutoSlider while we're dragging and apply it on mouse up.
BFC roots with children_are_inline()==true can still have floating boxes
as well. children_are_inline() is only concerned with in-flow children.
For this reason, we have to always consider floats when calculating
height:auto for BFC roots.
The SHA384 and SHA512 hashes would produce incorrect results for data
where the length % 128 was in the range 112-119. This was because the
total number of bits in the hashed values was added at the end as a
64-bit number instead of a 128-bit number. In most cases this would not
cause any issues, as this space was padded with zeroes, however in the
case that the length % 128 was 112-119, some incorrect data ended up
where this 128-bit length value was expected.
This change fixes the problems in LibTLS where some websites would
result in a DecryptError on handshake.
Rather than dividing the rect width and high by the border lengths,
this change multiples those lengths by the reciprocal of the width
and height because this is a faster operation. When mousing around on
the html spec website, the profile showed that inline painting
went from ~15% to ~3%
This fixes the placement of several background images on Acid2, most
notably the background of the eyes and the red rectangle near the bottom
of the head.
This is an editorial change in the Temporal spec.
See: https://github.com/tc39/proposal-temporal/commit/c5b645d
This means we now have to pass a global object and construct a BigInt
object just for the assertion, but oh well. We might want to have an
assertion macro that's optimized away in release builds at a later
point, however.
Collect all the preceding block-level siblings whose vertical margins
are collapsible. Both margin-top and margin-bottom now (previously,
we only considered the margin-bottom of siblings.)
Use the right margin in part-negative and all-negative situations.
We now distribute the line-height evenly between the space above and
below inline-level boxes. This noticeably improves our baseline
alignment in many cases.
Note that the "vertical-align: <length>" case is quite awkward, as the
extra height added by the offset baseline must count towards the line
box height.
There's a lot of room for improvement here, but this makes the buckets
container on Acid3 show up in the right place, with the right size.
Specifically HTMLIFrameElement and HTMLObjectElement. HTMLEmbedElement
will gain it automatically once it's also converted to inherit from
BrowsingContextContainer.
Make sure we use the create_anonymous_wrapper() helper function whenever
wrapping inline content in anonymous wrapper blocks. We were forgetting
to do this in one case, which led to some wrapper blocks having 0px
font-size and line-height.
Normally, paintable coordinates are relative to the nearest containing
block, but in the SVG case, since <svg> doesn't form a containing block,
we have to specialize the computation of SVGPaintable::absolute_rect().
When I wrote the An+B parser, it was guaranteed to have no
non-whitespace tokens after it. This is no longer true with the `of
foo` syntax, so this patch corrects the logic when there is no `+B`
segment.
This makes this case shown on Twitter work correctly. :^)
https://twitter.com/simevidas/status/1506657566012678151
We don't yet take the spread-distance parameter into account, since we
don't have a way to "inflate" the text shadow.
Also, I'm not sure if we need to inflate the shadow slightly anyway.
Blurred shadows of our pixel fonts seem very faint. Part of this is
that a blur of < 3px does nothing, see #13231, but even so we might
want to inflate it a little.
`text-shadow` does not support this, so this way we can still use the
same parsing code.
It's OK that we still assign a ShadowPlacement value to the
ShadowStyleValue, since it will just get ignored when painting
text-shadows, but if it appears in the property value then that is a
syntax error.
The `text-shadow` property is almost identical to `box-shadow`:
> Values are interpreted as for box-shadow [CSS-BACKGROUNDS-3].
> (But note that the inset keyword are not allowed.)
So, let's use the same data structures and parsing code for both. :^)
The HTMLObjectElement spec is set up to ignore application/octet-stream
MIME types only. For this to work, we need to implement the MIME type
sniffing algorithm so that all unknown MIME types become mapped to the
application/octet-stream type. Until then, ignore all application/ MIME
types as we won't be able to display them anyways.
We were using the literal string "unknown" as the unknown MIME type,
which caused us to treat the object as a nested browsing context (as
"unknown" does not start with "image/"). Use an Optional instead to
prevent this mishap.
This fixes a bug in the SQL REPL where after a user enters an
unrecognized command, the REPL would not print another "> " prompt and
would not accept any more input.
This converts the return value of File::read_link() from String to
ErrorOr<String>.
The rest of the change is to support the potential of an Error being
returned and subsequent release of the value when no Error is returned.
Unfortunately at this stage none of the places affected can utililize
our TRY() macro.
We currently only supported loading image data from an HTMLObjectElement
node. This adds (some) support for non-image data. A big FIXME is to
actually paint that data. We will need to make FrameBox and
NestedBrowsingContextPaintable work with HTMLObjectElement for this
(they currently only work with HTMLIFrameElement).
We will soon have two DOM nodes which contain nested browsing contexts:
HTMLIFrameElement and HTMLObjectElement. Only HTMLIFrameElement should
have its nested context created automatically upon insertion, so move
the invocation of that logic to HTMLIFrameElement.
HTMLObjectElement will need to be both a FormAssociatedElement and a
BrowsingContextContainer. Currently, both of these classes inherit from
HTMLElement. This can work in C++, but is generally frowned upon, and
doesn't play particularly well with the rest of LibWeb.
Instead, we can essentially revert commit 3bb5c62 to remove HTMLElement
from FormAssociatedElement's hierarchy. This means that objects such as
HTMLObjectElement individually inherit from FormAssociatedElement and
HTMLElement now.
Some caveats are:
* FormAssociatedElement still needs to know when the HTMLElement is
inserted into and removed from the DOM. This hook is automatically
injected via a macro now, while still allowing classes like
HTMLInputElement to also know when the element is inserted.
* Casting from a DOM::Element to a FormAssociatedElement is now a
sideways cast, rather than directly following an inheritance chain.
This means static_cast cannot be used here; but we can safely use
dynamic_cast since the only 2 instances of this already use RTTI to
verify the cast.
LibMain is dynamically linked in every binary. This results in a
slightly slower load time. In the past people have pegged this at 0.7
ms on some hardware.
This change makes it statically linked and eliminates 0.6 ms of
run-time on my machine. This is tested by running a script which just
executed `/bin/true` in a loop 10,000 times. Before this patch it
reliably executed in ~90,000 ms. After this patch it is ~84,000
ms. This is a speed up of 6,000 ms over 10,000 executions, or 0.6 ms
per execution.
This is another event upon which the task to determine an object's
respresentation must be queued:
* one of the element's ancestor object elements changes to or from
showing its fallback content
For example, on Acid2, the image for the eyes is nested below an object
that is designed to fail to load. This ensures the eyes will render as
the fallback of the failed parent object.
When a Resource is converted to an ImageResource, evict the original
resource from cache. The original resource's data has been moved, so on
a warm reload of a page, when that resource is loaded from cache, it
would not have any data to actually show.
Previously, if a user pressed Enter without typing a command at
the SQL REPL, the next line would be automatically indented. This
change makes it so we check if there were any tokens in the command
before applying the indentation logic.
For things like "line-height: 2" to work, the font size must be assigned
before resolving the line height. Previously, the line-height was
resolved first, which meant that numeric and other relative units were
resolved against the default font-size instead.
Our font database uses point sizes for fonts, and we were passing it
px sizes. This caused all fonts to be 1.333x larger than they should
be on the web. Of course it wasn't always noticeable with bitmap fonts,
but noticeable everywhere with scalable fonts.
There are a long list of conditions under which the HTMLObjectElement is
to queue an element task to load / determine an object's representation.
This handles the case where the data attribute has changed.
Much of the spec for determining the object's representation is not
implemented here. Namely, anything to do with XML documents or browser
plugins are left as FIXMEs.
HTMLObjectElement, when implemented according to the spec, does not know
the resource type specified by the 'data' attribute until after it has
actually loaded (i.e. it may be an image, XML document, etc.). Currently
we always use ImageLoader within HTMLObjectElement to load the object,
but will need to use ResourceLoader instead to generically load data.
However, ImageLoader / ImageResource have image-specific functionality
that HTMLObjectElement still needs if the resource turns out to be an
image. This patch will allow (only) HTMLObjectElement to convert the
generic Resource to an ImageResource as needed.
The spec at https://www.w3.org/TR/css-flexbox-1/ states that when
calculating specific spaces and sizes inside a flex container, the outer
size of a flex item needs to be taken into account.
This patch adds the margins in the main dimension of a flex item to
these calculations such that their margins are actually painted in a lot
of common cases.
It makes our Github page look marginally better.
This replaces the usage of `rounded_int_rect`, whose name did not
accurately reflect the rounding operation happening. For example, the
position of the rect was not rounded but floored, and the size was
pulled through `roundf` before casting to `int` which could result in
inadvertent flooring if the resulting floating point could not exactly
represent the rounded value.
There was an off-by-one bug in `Painter::do_draw_scaled_bitmap` where
the last column and row of the source bitmap would be skipped. This was
especially visible in PixelPaint when zooming in and out on smaller
images.
Instead of the top/left of the pixel, we now use the bottom/right side
of the pixel as a threshold to stop drawing.
If invoking a NodeFilter ends up deleting a node from the DOM, it's not
enough to only adjust the NodeIterator reference nodes in the
pre-removing steps. We must also adjust the current traversal pointer.
This is not in the spec, but it's how other engines behave, so let's do
the same.
I've encapsulated the Node + before-or-after-flag in a struct called
NodePointer so that we can use the same pre-removing steps for both the
traversal pointer and for the NodeIterator's reference node.
Note that when invoking the NodeFilter, we have to remember the node we
passed to the filter function, so that we can return it if accepted by
the filter.
This gets us another point on Acid3. :^)
The obsolete ttyname and ptsname syscalls are removed.
LibC doesn't rely on these anymore, and it helps simplifying the Kernel
in many places, so it's an overall an improvement.
In addition to that, /proc/PID/tty node is removed too as it is not
needed anymore by userspace to get the attached TTY of a process, as
/dev/tty (which is already a character device) represents that as well.
Instead, to determine these values (both the pts name and tty name), use
other methods. For determining the new name of the allocated psuedo
terminal, use ioctl on a file descriptor we got after opening /dev/ptmx
with the TIOCGPTN option.
For determining the name of TTY, we enumerate both /dev/pts and /dev
directories to find matching inode number and matching device mode.
This ioctl operation will allow userspace to determine the index number
of a MasterPTY after opening /dev/ptmx and actually getting an internal
file descriptor of MasterPTY.
CSS floats are now emitted by the InlineLevelIterator. When this
happens, IFC coordinates with the parent BFC to float the box to the
side, using the current LineBuilder state for vertical placement.
This makes the "instructions" text on Acid3 render as a single
contiguous flow of inline content.
This was implemented too rigidly, which made it impossible to place
floats correctly when they occurred in inline flow.
The new invariant is "all in-flow children must be either inline or
block". Out-of-flow children like floating and absolutely positioned
boxes are ignored when deciding when to generate anonymous boxes.