Bug 216145 – [css-grid] Incorrectly stretched SVGs without an aspect-ratio

• Status: NEW
• Component: CSS

[Oriol Brufau]

The following WPT tests are failing in WebKit:

css/css-grid/alignment/grid-item-no-aspect-ratio-stretch-5.html
css/css-grid/alignment/grid-item-no-aspect-ratio-stretch-6.html
css/css-grid/alignment/grid-item-no-aspect-ratio-stretch-7.html
css/css-grid/alignment/grid-item-no-aspect-ratio-stretch-8.html
css/css-grid/alignment/grid-item-no-aspect-ratio-stretch-9.html
css/css-grid/alignment/grid-item-no-aspect-ratio-stretch-10.html

https://wpt.fyi/results/css/css-grid/alignment?label=master&label=experimental&aligned&q=grid-item-no-aspect-ratio-stretch

Most of them (except 6 and 7) also fail in Chromium: https://crbug.com/1114013

[Radar WebKit Bug Importer]

<rdar://problem/68666896>

Created attachment 423644 [details]
Patch

[Javier Fernandez]

Comment on attachment 423644 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=423644&action=review

The patch looks good, but it doesn't fix all the tests that has been identified as failing due to this bug. We should either fix all of them or create a different bug, which this will depend on, and provide this patch as fix for that.

> Source/WebCore/ChangeLog:8
> +        When SVG element has fixed width or height as 0, it should also be reflected as true for has

s/for has/for

> Source/WebCore/ChangeLog:10
> +        RenderReplaced::computeReplacedLogicalHeight. This change is to differentiate the case when

Could you explain a bit more the case we want to differentiate from ?

> Source/WebCore/platform/graphics/FloatSize.h:75
> +    bool hasWidth = false;

Give than we set it to true if the Length value assigned to FloatSize attributes is Fixed or not, wouldn't be better to define these attributes as isFixed ?

> Source/WebCore/rendering/svg/RenderSVGRoot.cpp:83
> +    intrinsicSize.hasWidth = svgSVGElement().intrinsicWidth().isFixed();

The function computeIntrinsicRatioInformation is overridden by RenderImage as well; shouldn't this class set values to the hasWidth and hasHeight as well ? Otherwise, we'll use the default values for these FloatSize's fields in the RenderReplaced logic.

Additionally, are you sure that isFixed is enough ? Wouldn't be better to check use Length::isSpeficied() instead ? (it includes percent or calculated as well).

Fixes on 

css/css-grid/alignment/grid-item-no-aspect-ratio-stretch-6.html
css/css-grid/alignment/grid-item-no-aspect-ratio-stretch-7.html

are now at 

https://bugs.webkit.org/show_bug.cgi?id=223504

[Dawn Morningstar]

Created attachment 452396 [details]
expectations

[Dawn Morningstar]

*** Bug 236792 has been marked as a duplicate of this bug. ***

[Robert Jenner]

Comment on attachment 452396 [details]
expectations

Clearing flags on attachment: 452396

Committed r290038 (247416@trunk): <https://commits.webkit.org/247416@trunk>

[Dawn Morningstar]

(In reply to Robert Jenner from comment #7)
> Comment on attachment 452396 [details]
> expectations
> 
> Clearing flags on attachment: 452396
> 
> Committed r290038 (247416@trunk): <https://commits.webkit.org/247416@trunk>

I authored this commit, and Robert landed it for me, the purpose of the commit was to change the expectations on the two tests that did not receive changes, to relieve the EWS queue backlog.

My intention was only to fix 

css/css-grid/alignment/grid-item-no-aspect-ratio-stretch-6.html
css/css-grid/alignment/grid-item-no-aspect-ratio-stretch-7.html

and they are being addressed at bug 223504.