[8] Review request for 6847588: AWT test fails (original) (raw)

Sergey Bylokhov Sergey.Bylokhov at oracle.com
Tue Jun 11 03:53:46 PDT 2013

On 11.06.2013 14:01, Anthony Petrov wrote:

Anton: Your fix looks fine to me.

Sergey: I grepped our code for targetToPeer and getPeer and I see a lot more usages of the latter than the former. Could you please explain why using the targetToPeer is a correct way in this case? As I see it, it involves taking at least two locks in the AWTAutoShutdown code, and hence it seems kind of slow. Why would we prefer it over a direct call to (AWTAccessor.)getPeer? The difference is that SunToolkit.targetToPeer() works for trayIcon, Menu, MenuBar, MenuItem. So to make life easier I suggest everywhere use targetToPeer. And probably it is good idea to remove getPeer from AWTAccessor(17 usage in the jdk) -- best regards, Anthony On 06/10/2013 07:30 PM, Sergey Bylokhov wrote: Hi, Anton. I suppose the correct way to get peer from the component is to use (X/LWC)SunToolkit.targetToPeer() if possible.see latest version of LWKeyboardFocusManagerPeer.

On 10.06.2013 17:03, Anton Litvinov wrote: Hello,

Please review the following fix for the bug CR 6847588 which consists in calling of an improper implementation of the method "java.awt.Component.getPeer" in the class "sun.awt.X11.XKeyboardFocusManagerPeer". Webrev: http://cr.openjdk.java.net/~alitvinov/6847588/webrev.00 Thank you, Anton

-- Best regards, Sergey.

More information about the awt-dev mailing list