Bug 73683 – KeyframeValueList::operator= is neither safe to use nor prohibited

• Status: RESOLVED
• Resolution: FIXED
• Component: WebCore Misc.

[Yong Li]

When looking at GraphicLayer.h, I noticed that KeyframeValueList is copyable now, and it has a safe copy constructor. However, its default operator= is still unsafe.

We should either implement an operator= or simply disable it.

Not a big deal. But it would be nice we include something like this in webkit coding guidelines: when implement a copy constructor, please also implement operator=, and vice versa.

[Yong Li]

Created attachment 117669 [details]
make the operator= private

[Darin Adler]

Comment on attachment 117669 [details]
make the operator= private

This change is OK, but it would also be easy to write a swap and implement the canonical operator= instead.

[Darin Adler]

The canonical one is:

    KeyframeValueList& operator=(const KeyframeValueList& other)
    {
        KeyframeValueList copy(other);
        swap(copy);
        return *this;
    }

Assuming we write a swap function that swaps each member.

[Yong Li]

Created attachment 117874 [details]
Implement operator= with the copy constructor.

How about this? :)

[Gyuyoung Kim]

Comment on attachment 117874 [details]
Implement operator= with the copy constructor.

Attachment 117874 [details] did not pass efl-ews (efl):
Output: http://queues.webkit.org/results/10732638

[Early Warning System Bot]

Comment on attachment 117874 [details]
Implement operator= with the copy constructor.

Attachment 117874 [details] did not pass qt-ews (qt):
Output: http://queues.webkit.org/results/10736319

[WebKit Review Bot]

Comment on attachment 117874 [details]
Implement operator= with the copy constructor.

Attachment 117874 [details] did not pass chromium-ews (chromium-xvfb):
Output: http://queues.webkit.org/results/10734669

[Yong Li]

Created attachment 117892 [details]
impl operator= with copy constructor

[Darin Adler]

Comment on attachment 117892 [details]
impl operator= with copy constructor

View in context: https://bugs.webkit.org/attachment.cgi?id=117892&action=review

> Source/WebCore/platform/graphics/GraphicsLayer.h:182
> +    KeyframeValueList& operator=(const KeyframeValueList& o)
> +    {
> +        if (this == &o)
> +            return *this;
> +
> +        this->~KeyframeValueList();
> +        new (this) KeyframeValueList(o);
> +        return *this;
> +    }

The copy and swap idiom is usually preferred over this idiom. I’d like to use that instead.

[Yong Li]

(In reply to comment #9)
> (From update of attachment 117892 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=117892&action=review
> 
> > Source/WebCore/platform/graphics/GraphicsLayer.h:182
> > +    KeyframeValueList& operator=(const KeyframeValueList& o)
> > +    {
> > +        if (this == &o)
> > +            return *this;
> > +
> > +        this->~KeyframeValueList();
> > +        new (this) KeyframeValueList(o);
> > +        return *this;
> > +    }
> 
> The copy and swap idiom is usually preferred over this idiom. I’d like to use that instead.

Copy and swap performs one more copy if not optimized by compiler, and it also needs a swap method. It is first time I got the dtor/copy ctor idiom when writing operator=, and I love it immediately. I even hope it could be a standard that all c++ compilers should implicitly implement operator= in this way when a copy ctor is available: destruct the object and copy-construct a new one on the zombie.

One of the pros of the copy and swap idiom is it doesn't need to check (this == &o) assuming swap() already checks that.

Also, I cannot find any "new (this)" in current webkit code, which may a little bit tricky. So I'll upload a copy&swap version.

[WebKit Review Bot]

Comment on attachment 117892 [details]
impl operator= with copy constructor

Attachment 117892 [details] did not pass chromium-ews (chromium-xvfb):
Output: http://queues.webkit.org/results/10727906

New failing tests:
svg/custom/linking-uri-01-b.svg

[Yong Li]

Created attachment 117936 [details]
implement operator= with swap()

[WebKit Review Bot]

Comment on attachment 117936 [details]
implement operator= with swap()

Clearing flags on attachment: 117936

Committed r102092: <http://trac.webkit.org/changeset/102092>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.