Bug 128671 – CounterContentData::counter() and ImageContentData::image() should return references

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

[David Kilzer (:ddkilzer)]

CounterContentData::counter() and ImageContentData::image() should return references.

Follow-up to Darin's comment in Bug 128538 Comment #3:

(In reply to bug 128538 comment #3)
> (From update of attachment 223726 [details])
> > Source/WebCore/rendering/style/ContentData.h:160
> >      const CounterContent* counter() const { return m_counter.get(); }
> 
> Seems like this should return a reference rather than a pointer, unless it can be null.

[David Kilzer (:ddkilzer)]

Created attachment 223975 [details]
Patch v1

[Andreas Kling]

Comment on attachment 223975 [details]
Patch v1

View in context: https://bugs.webkit.org/attachment.cgi?id=223975&action=review

> Source/WebCore/rendering/style/ContentData.h:94
>      void setImage(PassRefPtr<StyleImage> image) { m_image = image; }

Is this function ever used in a way that makes m_image null?

[Darin Adler]

Comment on attachment 223975 [details]
Patch v1

View in context: https://bugs.webkit.org/attachment.cgi?id=223975&action=review

> Source/WebCore/css/CSSComputedStyleDeclaration.cpp:1408
> +            const CounterContent& counter = toCounterContentData(contentData)->counter();
> +            list.get().append(cssValuePool().createValue(counter.identifier(), CSSPrimitiveValue::CSS_COUNTER_NAME));

Why not auto&?

Maybe drop the local variable entirely and do this in a single line?

> Source/WebCore/css/CSSComputedStyleDeclaration.cpp:1411
> +            const StyleImage& image = toImageContentData(contentData)->image();
> +            list.get().append(image.cssValue());

Why not auto&?

Maybe drop the local variable entirely and do this in a single line?

> Source/WebCore/css/StyleResolver.cpp:3552
> +                        RefPtr<StyleImage> loadedImage = loadPendingImage(static_cast<StylePendingImage*>(const_cast<StyleImage*>(&image)));

This const cast is ugly.

Separate point: Can we add checked casts for StyleImage?

> Source/WebCore/rendering/RenderElement.cpp:135
> +        const StyleImage& styleImage = toImageContentData(contentData)->image();

Why not auto&?

> Source/WebCore/rendering/style/ContentData.h:93
> +    const StyleImage& image() const { return *m_image; }

Why did you get rid of the const overload?

>> Source/WebCore/rendering/style/ContentData.h:94
>>      void setImage(PassRefPtr<StyleImage> image) { m_image = image; }
> 
> Is this function ever used in a way that makes m_image null?

Probably should add an assertion here too, as in the constructor and consider changing to PassRef maybe?

> Source/WebCore/rendering/style/ContentData.h:156
>      }

We should assert m_counter.

> Source/WebCore/rendering/style/ContentData.h:159
>      void setCounter(std::unique_ptr<CounterContent> counter) { m_counter = std::move(counter); }

We should assert counter (before) or m_counter (after).

[David Kilzer (:ddkilzer)]

(In reply to comment #3)
> (From update of attachment 223975 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=223975&action=review
> 
> > Source/WebCore/css/CSSComputedStyleDeclaration.cpp:1408
> > +            const CounterContent& counter = toCounterContentData(contentData)->counter();
> > +            list.get().append(cssValuePool().createValue(counter.identifier(), CSSPrimitiveValue::CSS_COUNTER_NAME));
> 
> Why not auto&?

Ahh, I didn't realize one had to use 'auto&' for a reference.  I'll switch them back.

> Maybe drop the local variable entirely and do this in a single line?

I don't think it improves readability to make this long line even longer.

> > Source/WebCore/css/CSSComputedStyleDeclaration.cpp:1411
> > +            const StyleImage& image = toImageContentData(contentData)->image();
> > +            list.get().append(image.cssValue());
> 
> Why not auto&?

Will fix.

> Maybe drop the local variable entirely and do this in a single line?

This line still seems readable as one line.

> > Source/WebCore/css/StyleResolver.cpp:3552
> > +                        RefPtr<StyleImage> loadedImage = loadPendingImage(static_cast<StylePendingImage*>(const_cast<StyleImage*>(&image)));
> 
> This const cast is ugly.

It gets cleaned up in a subsequent patch that makes loadPendingImage() take a reference.

> Separate point: Can we add checked casts for StyleImage?

I have a follow-up patch to add the checked casts for StyleImage, but it currently depends on this patch.

> > Source/WebCore/rendering/RenderElement.cpp:135
> > +        const StyleImage& styleImage = toImageContentData(contentData)->image();
> 
> Why not auto&?

Will fix.

> > Source/WebCore/rendering/style/ContentData.h:93
> > +    const StyleImage& image() const { return *m_image; }
> 
> Why did you get rid of the const overload?

It didn't really add anything.  (As noted above, the const cast goes away in a future patch.)

> >> Source/WebCore/rendering/style/ContentData.h:94
> >>      void setImage(PassRefPtr<StyleImage> image) { m_image = image; }
> > 
> > Is this function ever used in a way that makes m_image null?
> 
> Probably should add an assertion here too, as in the constructor and consider changing to PassRef maybe?

Good point.  Will add.

> > Source/WebCore/rendering/style/ContentData.h:156
> >      }
> 
> We should assert m_counter.

Will add.

> > Source/WebCore/rendering/style/ContentData.h:159
> >      void setCounter(std::unique_ptr<CounterContent> counter) { m_counter = std::move(counter); }
> 
> We should assert counter (before) or m_counter (after).

Will add.

[David Kilzer (:ddkilzer)]

Committed r164224: <http://trac.webkit.org/changeset/164224>

[David Kilzer (:ddkilzer)]

(In reply to comment #4)
> (In reply to comment #3)
> > Separate point: Can we add checked casts for StyleImage?
> 
> I have a follow-up patch to add the checked casts for StyleImage, but it currently depends on this patch.

Bug 128915

[David Kilzer (:ddkilzer)]

(In reply to comment #4)
> (In reply to comment #3)
> > (From update of attachment 223975 [details] [details])
> > View in context: https://bugs.webkit.org/attachment.cgi?id=223975&action=review
> > 
> > > Source/WebCore/css/StyleResolver.cpp:3552
> > > +                        RefPtr<StyleImage> loadedImage = loadPendingImage(static_cast<StylePendingImage*>(const_cast<StyleImage*>(&image)));
> > 
> > This const cast is ugly.
> 
> It gets cleaned up in a subsequent patch that makes loadPendingImage() take a reference.

Bug 129021