Bug 178640 – [iOS] WebProcess::initializeWebProcess spends ~150ms spinning up AVSystemController on some devices

• Component: WebRTC
• Status: RESOLVED FIXED

[Wenson Hsieh]

Optimizing this would improve web process bring-up time on certain devices.

[Wenson Hsieh]

<rdar://problem/35113105>

[Wenson Hsieh]

Created attachment 324517 [details]
Patch

[youenn fablet]

Comment on attachment 324517 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=324517&action=review

> Source/WebKit/WebProcess/WebPage/ios/WebPageIOS.mm:3421
> +        NSError *error = nil;

Is is usually style for mm file to do "NSError *error" in lieu of CPP like "NSError* error"?

[Wenson Hsieh]

Comment on attachment 324517 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=324517&action=review

Thanks for the review!

>> Source/WebKit/WebProcess/WebPage/ios/WebPageIOS.mm:3421
>> +        NSError *error = nil;
> 
> Is is usually style for mm file to do "NSError *error" in lieu of CPP like "NSError* error"?

Correct. It's this rule, in the WebKit style guidelines: <https://webkit.org/code-style-guidelines/#pointers-non-cpp>. Although, it seems a common interpretation of this rule (and an interpretation I personally find more useful :)) is that Objective C objects should have a space between the type and the *, while pointers to C++ and C classes and structs should not have an extra space. FWIW, this appears to be the convention followed here, in WebPageIOS.mm, as well as many other Objective-C++ files in WebKit (e.g. search for "FrameView*" in this file).

[WebKit Commit Bot]

Comment on attachment 324517 [details]
Patch

Clearing flags on attachment: 324517

Committed r223827: <https://trac.webkit.org/changeset/223827>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.