This is to allow running Ladybird without the SQL database for testing.
Primarily, this will let us set 'follow-fork-mode' to 'child' within GDB
to enter the WebContent process, rather than the SQLServer process. But
this is also handy for digging into cookie storage issues.
This is to allow testing autoplay, poster images, etc. without having to
stash local changes to the page. This also changes the URLs used on the
page to be relative to the page itself, to allow the page to load both
on Serenity and Lagom.
This will fetch the URL indicated by the poster attribute when it's set,
changed, or removed. The spec doesn't say how to handle animated poster
images, so we just grab the first frame of the image, which seems to
match other implementations.
This will be needed by the layout node, which may change what is painted
when the position of the frame image is not the same as the element's
current time.
This will be needed by the layout node, which may change what is painted
when the position of the frame image is not the same as the element's
current time.
It returns a PaintableBox (a PaintableWithLines, to be specific), not a
'PaintBox'. paintable_box() without the cast is already available
through BlockContainer's Box base class, we don't need to shadow it.
The GUI now tracks when it becomes disconnected from ChessEngine.
If not currently waiting for a move from ChessEngine, it will
automatically reconnect on the next engine move. If a disconnection
occurs while waiting for a move, the player is asked whether they
want to try again or not.
We are currently using the fetch controller's terminate() method to stop
ongoing fetches when the HTMLMediaElement load algorithm is invoked.
This method ultimately causes the fetch response to be a network error,
which we propagate through the HTMLMediaElement's error event. This can
cause websites, such as Steam, to avoid attempting to play any video.
The spec does not actually specify what it means to "stop" or "cancel" a
fetching process. But we should not use terminate() as that is a defined
spec method, and the spec does tend to indicate when that method should
be used (e.g. as it does in XMLHttpRequest).
The HTMLMediaElement will need to stop fetching processes when its load
algorithm is invoked while a fetch is ongoing. We don't have a way to
really stop the process, due to the way it runs on nested deferred task
invocations. So for now, this swaps the fetch callbacks (e.g. to process
a fetch response) with empty callbacks.
This is the first step towards implementing the new "navigable" concept
from the HTML spec.
Co-authored-by: Aliaksandr Kalenik <kalenik.aliaksandr@gmail.com>
The "browsing context container" concept in the HTML spec has been
replaced with "navigable container". Renaming this is the first step of
many towards implementing the new world.
Co-authored-by: Aliaksandr Kalenik <kalenik.aliaksandr@gmail.com>
Now that LengthStyleValue never contains `auto`, IdentifierStyleValue is
the only type that can hold an identifier. This lets us remove a couple
of virtual methods from StyleValue.
I've kept `has_auto()` and `to_identifier()` for convenience, but they
are now simple non-virtual methods.
Previously, whether trying to parse a `<length>` or `<dimension>`, we
would accept `auto` and produce a `LengthStyleValue` from it. This
would fool the `property_accepts_value()` into allowing `auto` where it
does not belong, if the property did accept lengths.
Of the few places in the parser that called `parse_dimension_value()` or
`parse_length()`, none of them were expecting it to accept `auto`, so
this fixes those too. :^)
Currently, `property_accepts_value()` always returns `true` if the
property is a shorthand. (This is a bug, and should actually be fixed
properly at some point...) This means that all identifiers are caught
here, including `auto`, which should be handled by the `flex-basis`
branch instead.
This works currently because `auto` is a LengthStyleValue, but that's
about to change!
