[Webkit-unassigned] [Bug 167415] Avoid evicting link preload resources when parsing is done.

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=167415

--- Comment #18 from Yoav Weiss <yoav at yoav.ws> ---
(In reply to comment #17)
> Comment on attachment 299800 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=299800&action=review
> 
> > Source/WebCore/dom/Document.cpp:5046
> > +    // Parser should have picked up all speculative preloads by now
> 
> Is there a difference between a speculative preload and a preload?

A speculative preload is one issued by the preloadScanner (at least that's the terminology that I used)

> 
> > Source/WebCore/loader/LinkLoader.cpp:183
> >      linkRequest.setInitiator("link");
> > +    linkRequest.setIsLinkPreload();
> 
> Do we really need another bool for this?  Can't we just check if the
> initiator is "link"?

A "link" initiator can also be a stylesheet. A slightly similar flag is the ForPreload flag, which signified if a request and a resource were triggered as a speculative preload.

Maybe we can somehow unite those flags, but a 4 state enum is not necessarily better/clearer than two bools

> 
> > Source/WebCore/loader/LinkPreloadResourceClients.h:66
> > +            m_resource->cancelLoad();
> >              m_resource->removeClient(client);
> 
> I think removing the client will cancel the load if this was the only
> client.  If it wasn't the only client, then something else has made the same
> request and we probably don't want to cancel the load for that other client,
> right?

I'll try that

> 
> > Source/WebCore/loader/cache/CachedResourceLoader.h:131
> > +    enum ClearPreloadsMode { ClearSpeculativePreloads, ClearAllPreloads };
> 
> enum class

OK

> 
> > Source/WebCore/loader/cache/CachedResourceRequest.cpp:48
> > +    , m_isLinkPreload(false)
> 
> If we added a bool here, we would want to have its initializer be an
> initializer list in the header
> bool m_isLinkPreload { false };

OK

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.webkit.org/pipermail/webkit-unassigned/attachments/20170126/4b148241/attachment.html>