[webkit-reviews] review denied: [Bug 49735] [Chromium] Introduce wrapper types for WebCore::GeolocationError, WebCore::GeolocationPosition, WebCore::GeolocationController : [Attachment 74382] Patch

[bugzilla-daemon at webkit.org]

Darin Fisher (:fishd, Google) <fishd at chromium.org> has denied John Knottenbelt
<jknotten at chromium.org>'s request for review:
Bug 49735: [Chromium] Introduce wrapper types for WebCore::GeolocationError,
WebCore::GeolocationPosition, WebCore::GeolocationController
https://bugs.webkit.org/show_bug.cgi?id=49735

Attachment 74382: Patch
https://bugs.webkit.org/attachment.cgi?id=74382&action=review

------- Additional Comments from Darin Fisher (:fishd, Google)
<fishd at chromium.org>
View in context: https://bugs.webkit.org/attachment.cgi?id=74382&action=review

> WebKit/chromium/public/WebGeolocationController.h:59
> +    WebCore::GeolocationController* m_controller;

what prevents the GeolocationController object from being destroyed before the
WebGeolocationController?
how do you avoid memory errors in that case?

this seems error prone.

> WebKit/chromium/public/WebGeolocationError.h:38
> +class WebGeolocationError {

GeolocationError is reference counted, so it seems like you should implement
WebGeolocationError as a simple wrapper (like WebNode).

Use WebPrivatePtr<WebCore::GeolocationError> as your sole member variable.

> WebKit/chromium/public/WebGeolocationPosition.h:88
> +    void copyFrom(const WebCore::GeolocationPosition&);

GeolocationPosition is reference counted, so it seems like you should implement

WebGeolocationPosition as a simple wrapper (like WebNode).

Use WebPrivatePtr<WebCore::GeolocationPosition> as your sole member variable.