Bug 95036 – Prerenderering should gracefully handle no PrerendererClient or PrerenderingPlatform being provided.

• Status: RESOLVED
• Resolution: FIXED
• Component: New Bugs

[Gavin Peters]

Prerenderering should gracefully handle no PrerendererClient or PrerenderingPlatform being provided.

[Gavin Peters]

Created attachment 160610 [details]
Patch

[Gavin Peters]

Comment on attachment 160610 [details]
Patch

abarth, WDYT?

[Adam Barth]

Comment on attachment 160610 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=160610&action=review

> Source/WebCore/loader/Prerenderer.cpp:140
> -    if (!m_client)
> +    if (!m_initializedClient) {
> +        m_initializedClient = true;
>          m_client = PrerendererClient::from(document()->page());
> +    }
>      return m_client;

Should we just do this in the constructor?  I don't remember the exactly access sequence for these objects.

[Gavin Peters]

(In reply to comment #3)
> (From update of attachment 160610 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=160610&action=review
> 
> > Source/WebCore/loader/Prerenderer.cpp:140
> > -    if (!m_client)
> > +    if (!m_initializedClient) {
> > +        m_initializedClient = true;
> >          m_client = PrerendererClient::from(document()->page());
> > +    }
> >      return m_client;
> 
> Should we just do this in the constructor?  I don't remember the exactly access sequence for these objects.

The problem is that the client isn't provided until that platform initializes, which can be later. When I did it in the constructor, I had crashes.

So I think this logic at a minimum needs to be commented.

[Gavin Peters]

Created attachment 160615 [details]
Patch

[WebKit Review Bot]

Comment on attachment 160615 [details]
Patch

Clearing flags on attachment: 160615

Committed r126714: <http://trac.webkit.org/changeset/126714>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.