Simplify process suppression state calculation

[Gavin Barraclough]

Don't check the application occlusion state (this is covered by the pages already), and let the page check visually idle. Remove layers of functions.

[Gavin Barraclough]

Created attachment 220358 [details]
Fix

[Geoffrey Garen]

Comment on attachment 220358 [details]
Fix

View in context: https://bugs.webkit.org/attachment.cgi?id=220358&action=review

> ../OpenSource/Source/WebKit2/UIProcess/WebPageProxy.cpp:520
> +    return m_viewState & ViewState::IsVisuallyIdle && m_pageGroup->preferences()->pageVisibilityBasedProcessSuppressionEnabled();

Feels like this should have parens to indicate associativity.

> ../OpenSource/Source/WebKit2/UIProcess/mac/WebContextMac.mm:391
> +    for (const auto& context : WebContext::allContexts()) {

Why does this loop use "const auto&" while the other uses "auto*"?

[Geoffrey Garen]

Comment on attachment 220358 [details]
Fix

Since allContexts() is a vector of WebContext*, and since the range-based for loops do not intend to modify the contents of the vector, I think they should both just say "auto" -- neither "auto*" nor "const auto&".

[Sam Weinig]

Comment on attachment 220358 [details]
Fix

View in context: https://bugs.webkit.org/attachment.cgi?id=220358&action=review

r=me as well.

>> ../OpenSource/Source/WebKit2/UIProcess/mac/WebContextMac.mm:391
>> +    for (const auto& context : WebContext::allContexts()) {
> 
> Why does this loop use "const auto&" while the other uses "auto*"?

This should use auto*.

[Gavin Barraclough]

Committed revision 161301.