Bug 233109 – Implement parsing and animation support for offset shorthand

• Status: RESOLVED
• Resolution: FIXED
• Component: CSS

[Kiet Ho]

Implement parsing and animation support for offset shorthand

[Kiet Ho]

Created attachment 444318 [details]
WIP

[EWS Watchlist]

This patch modifies the imported WPT tests. Please ensure that any changes on the tests (not coming from a WPT import) are exported to WPT. Please see https://trac.webkit.org/wiki/WPTExportProcess

[Kiet Ho]

*** Bug 232760 has been marked as a duplicate of this bug. ***

[Kiet Ho]

Created attachment 444324 [details]
WIP

[Kiet Ho]

Created attachment 444836 [details]
WIP

[Kiet Ho]

Still needs to figure out what is the correct serialization format. In particular, is it needed to make the serialization as short as possible?

For example, Chrome and this patch serializes "auto none" to "auto none". However, "auto" is equivalent and shorter.

On the other hand, "auto none reverse" must be serialized to "auto none reverse" and not "auto reverse", because according to the grammar, if offset-rotate is serialized, then so is offset-path.

[Radar WebKit Bug Importer]

<rdar://problem/85611877>

[Nikos Mouchtaris]

Created attachment 451649 [details]
Patch

[Nikos Mouchtaris]

Created attachment 451727 [details]
Patch

[Dean Jackson]

Comment on attachment 451727 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=451727&action=review

> Source/WebCore/css/CSSComputedStyleDeclaration.cpp:2640
> +    {
> +        auto innerList = CSSValueList::createSpaceSeparated();
> +        innerList->append(valueForPositionOrAuto(style, style.offsetPosition()));
> +        innerList->append(valueForPathOperation(style, style.offsetPath(), SVGPathConversion::ForceAbsolute));
> +        innerList->append(CSSValuePool::singleton().createValue(style.offsetDistance(), style));
> +        innerList->append(valueForOffsetRotate(style.offsetRotate()));
> +
> +        outerList->append(WTFMove(innerList));
> +    }

Any reason for this being in its own  scope?

> Source/WebCore/css/StyleProperties.cpp:466
> +static bool isAutoKeyword(const CSSValue& value)
> +{
> +    if (!is<CSSPrimitiveValue>(value))
> +        return false;
> +
> +    auto& primitiveValue = downcast<CSSPrimitiveValue>(value);
> +    return primitiveValue.isValueID() && primitiveValue.valueID() == CSSValueAuto;
> +}

Is this really the first/only time we're checking for this keyword in the file?

> Source/WebCore/css/StyleProperties.cpp:508
> +            if (!offsetDistance.value() || !is<CSSPrimitiveValue>(offsetDistance.value()))

Maybe a local variable for offsetDistance.value() to avoid calling it multiple times?

> Source/WebCore/css/StyleProperties.cpp:527
> +            if (!offsetRotate.value() || !is<CSSOffsetRotateValue>(offsetRotate.value()))

Same with this .value()

[Nikos Mouchtaris]

Created attachment 452096 [details]
Patch

[Nikos Mouchtaris]

(In reply to Dean Jackson from comment #10)
> Comment on attachment 451727 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=451727&action=review
> 
> > Source/WebCore/css/CSSComputedStyleDeclaration.cpp:2640
> > +    {
> > +        auto innerList = CSSValueList::createSpaceSeparated();
> > +        innerList->append(valueForPositionOrAuto(style, style.offsetPosition()));
> > +        innerList->append(valueForPathOperation(style, style.offsetPath(), SVGPathConversion::ForceAbsolute));
> > +        innerList->append(CSSValuePool::singleton().createValue(style.offsetDistance(), style));
> > +        innerList->append(valueForOffsetRotate(style.offsetRotate()));
> > +
> > +        outerList->append(WTFMove(innerList));
> > +    }
> 
> Any reason for this being in its own  scope?
Fixed.
> 
> > Source/WebCore/css/StyleProperties.cpp:466
> > +static bool isAutoKeyword(const CSSValue& value)
> > +{
> > +    if (!is<CSSPrimitiveValue>(value))
> > +        return false;
> > +
> > +    auto& primitiveValue = downcast<CSSPrimitiveValue>(value);
> > +    return primitiveValue.isValueID() && primitiveValue.valueID() == CSSValueAuto;
> > +}
> 
> Is this really the first/only time we're checking for this keyword in the
> file?
No reason for this to be its own function.
> 
> > Source/WebCore/css/StyleProperties.cpp:508
> > +            if (!offsetDistance.value() || !is<CSSPrimitiveValue>(offsetDistance.value()))
> 
> Maybe a local variable for offsetDistance.value() to avoid calling it
> multiple times?
Fixed.
> 
> > Source/WebCore/css/StyleProperties.cpp:527
> > +            if (!offsetRotate.value() || !is<CSSOffsetRotateValue>(offsetRotate.value()))
> 
> Same with this .value()
Fixed.

[Nikos Mouchtaris]

Created attachment 452109 [details]
Patch

[EWS]

Committed r289876 (247314@main): <https://commits.webkit.org/247314@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 452109 [details].