REGRESSION (r254406): Gmail.com star/favorite icons are not rendering

• Component: Images
• Status: RESOLVED FIXED

[Simon Fraser (smfr)]

r254406 broke the icons in the "star" column in Gmail.

The ::before has:
content: -webkit-image-set('' 1x,'' 2x);

It seems this was previously invalid, so that content:'' was used. But now it's treated as valid, so we show the missing image icon.

I think we need to preserve the old behavior of image-set in -webkit-image-set.

[Simon Fraser (smfr)]

rdar://problem/58858225

[Noam Rosenthal]

(In reply to Simon Fraser (smfr) from comment #0)
> r254406 broke the icons in the "star" column in Gmail.
> 
> The ::before has:
> content: -webkit-image-set('' 1x,'' 2x);
> 
> It seems this was previously invalid, so that content:'' was used. But now
> it's treated as valid, so we show the missing image icon.
> 
> I think we need to preserve the old behavior of image-set in
> -webkit-image-set.

or maybe disallow empty images?

[Noam Rosenthal]

(In reply to Noam Rosenthal from comment #2)
> (In reply to Simon Fraser (smfr) from comment #0)
> > r254406 broke the icons in the "star" column in Gmail.
> > 
> > The ::before has:
> > content: -webkit-image-set('' 1x,'' 2x);
> > 
> > It seems this was previously invalid, so that content:'' was used. But now
> > it's treated as valid, so we show the missing image icon.
> > 
> > I think we need to preserve the old behavior of image-set in
> > -webkit-image-set.
> 
> or maybe disallow empty images?
I think I'll do both for now.

[Noam Rosenthal]

Created attachment 389111 [details]
Patch

[Simon Fraser (smfr)]

(In reply to Noam Rosenthal from comment #2)
> (In reply to Simon Fraser (smfr) from comment #0)
> > r254406 broke the icons in the "star" column in Gmail.
> > 
> > The ::before has:
> > content: -webkit-image-set('' 1x,'' 2x);
> > 
> > It seems this was previously invalid, so that content:'' was used. But now
> > it's treated as valid, so we show the missing image icon.
> > 
> > I think we need to preserve the old behavior of image-set in
> > -webkit-image-set.
> 
> or maybe disallow empty images?

Are empty images allowed per spec?

[Simon Fraser (smfr)]

Comment on attachment 389111 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=389111&action=review

> Source/WebCore/css/parser/CSSPropertyParserHelpers.cpp:1706
> +    if ((range.peek().type() == StringToken) && (allowedImageTypes & AllowedImageType::RawStringAsURL)) {

I think we prefer allowedImageTypes.contains(AllowedImageType::RawStringAsURL)

> Source/WebCore/css/parser/CSSPropertyParserHelpers.cpp:1714
>          CSSValueID id = range.peek().functionId();

Not this patch, but please don't use 'id'; it's a reserved word in Objective-C.

> Source/WebCore/css/parser/CSSPropertyParserHelpers.cpp:1715
> +        if ((allowedImageTypes & AllowedImageType::GeneratedImage) && isGeneratedImage(id))

.contains()

> Source/WebCore/css/parser/CSSPropertyParserHelpers.cpp:1718
> +        if (allowedImageTypes & AllowedImageType::ImageSet) {

.contains()

> Source/WebCore/css/parser/CSSPropertyParserHelpers.cpp:1726
> +    if (allowedImageTypes & AllowedImageType::URLFunction) {

.contains()

[Noam Rosenthal]

Created attachment 389238 [details]
Patch for landing

[Noam Rosenthal]

(In reply to Simon Fraser (smfr) from comment #5)
> (In reply to Noam Rosenthal from comment #2)
> > (In reply to Simon Fraser (smfr) from comment #0)
> > > r254406 broke the icons in the "star" column in Gmail.
> > > 
> > > The ::before has:
> > > content: -webkit-image-set('' 1x,'' 2x);
> > > 
> > > It seems this was previously invalid, so that content:'' was used. But now
> > > it's treated as valid, so we show the missing image icon.
> > > 
> > > I think we need to preserve the old behavior of image-set in
> > > -webkit-image-set.
> > 
> > or maybe disallow empty images?
> 
> Are empty images allowed per spec?

They actually are, considered empty URLs. Removed that part from the patch.

[WebKit Commit Bot]

Comment on attachment 389238 [details]
Patch for landing

Clearing flags on attachment: 389238

Committed r255420: <https://trac.webkit.org/changeset/255420>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Darin Adler]

Did we check the behavior of other browsers in image-set? It would be unfortunate if we were the only one who matched the spec on this.

[Noam Rosenthal]

(In reply to Darin Adler from comment #11)
> Did we check the behavior of other browsers in image-set? It would be
> unfortunate if we were the only one who matched the spec on this.

Of course, working on other browsers in parallel, currently getting the W3C tests around this to be comprehensive.