[webkit-reviews] review denied: [Bug 54969] [Chromium] Implement WebKit methods to assist with Cocoa NSTextInput implementation : [Attachment 86977] Patch

[bugzilla-daemon at webkit.org]

Alexey Proskuryakov <ap at webkit.org> has denied Robert Sesek
<rsesek at chromium.org>'s request for review:
Bug 54969: [Chromium] Implement WebKit methods to assist with Cocoa NSTextInput
implementation
https://bugs.webkit.org/show_bug.cgi?id=54969

Attachment 86977: Patch
https://bugs.webkit.org/attachment.cgi?id=86977&action=review

------- Additional Comments from Alexey Proskuryakov <ap at webkit.org>
View in context: https://bugs.webkit.org/attachment.cgi?id=86977&action=review

This will break Mac build, because the new functions need to be added to
WebCore.base.exp.

Looks like a good improvement in cross platform code. r- for build breakage,
and for many comments from the reviewers.

> Source/WebKit/chromium/src/mac/WebTextHelper.mm:129
> +    NSData* archivedData([NSArchiver archivedDataWithRootObject:string]);

I'm not sure if using NSArchiver is a good idea. Are you archiving in renderer
process, and unarchiving in main one? If the renderer is compromised, it could
send anything (like an NSWindow) in the serialized data.

> Source/WebKit/mac/WebView/WebFrame.mm:676
> +    if (range && TextIterator::locationAndLengthFromRange(range, location,
length))
> +	   return NSMakeRange(location, length);
> +    return NSMakeRange(NSNotFound, 0);

We generally prefer early return on error. It's more readable when the main
code path is linear.

> Source/WebKit2/WebProcess/WebPage/mac/WebPageMac.mm:164
> +    size_t loc, len;

Please don't abbreviate.