Use new RegistrableDomain type in PSON code

[Chris Dumez]

Use new RegistrableDomain type in PSON code instead of more error-prone String type.

[Chris Dumez]

Created attachment 364425 [details]
Patch

[John Wilander]

Comment on attachment 364425 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=364425&action=review

> Source/WebKit/UIProcess/SuspendedPageProxy.cpp:85
> +    , m_registrableDomain(URL(URL(), item.url()))

Do you need to construct a new URL with item.url()?

[Chris Dumez]

Comment on attachment 364425 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=364425&action=review

>> Source/WebKit/UIProcess/SuspendedPageProxy.cpp:85
>> +    , m_registrableDomain(URL(URL(), item.url()))
> 
> Do you need to construct a new URL with item.url()?

item.url() returns a String, not a URL.

[Chris Dumez]

ping review?

[youenn fablet]

Comment on attachment 364425 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=364425&action=review

> Source/WebKit/UIProcess/WebProcessProxy.h:118
> +    WebCore::RegistrableDomain registrableDomain() const { return m_registrableDomain.valueOr(WebCore::RegistrableDomain { }); }

It seems we can have m_registrableDomain being nullopt, isEmpty or being a real registrable domain.
But we do not expose the difference between nullopt and isEmpty in the public getter.
Can it be a potential source of issues in the future?

[Chris Dumez]

Comment on attachment 364425 [details]
Patch

Clearing flags on attachment: 364425

Committed r242889: <https://trac.webkit.org/changeset/242889>

[Chris Dumez]

All reviewed patches have been landed.  Closing bug.

[Radar WebKit Bug Importer]

<rdar://problem/48853537>