Ok, then fix looks fine.
On 19.09.2013 15:00, Leonid Romanov wrote:
I didn't want to alter existing Toolkit.getProperty() logic: currently, calling
Toolkit.getProperty() won't trigger platform's toolkit initialization. My fix
keeps this logic intact. Implementing your suggestion will break it. I'm afraid
that this might break some things somewhere unexpectedly.
On 19.09.2013, at 14:35, Sergey Bylokhov <sergey.bylok...@oracle.com> wrote:
Hello,
Why not initialize default toolkit via "getDefaultToolkit!=this" explicitly?
On 13.09.2013 17:19, Anthony Petrov wrote:
Thanks for the updated fix. Looks fine to me.
--
best regards,
Anthony
On 09/13/2013 05:10 PM, Leonid Romanov wrote:
The main bug simply states that "accelerators are displayed as Meta instead of the
Command symbol", but if you look at duplicates they are mostly concerned about
displaying correct key stroke names in tooltips. So, I think on practice we can expect
that LWCToolkit have been initialized. Here is an updated fix:
http://cr.openjdk.java.net/~leonidr/7129133/webrev.01/
On Sep 13, 2013, at 4:12 PM, Anthony Petrov <anthony.pet...@oracle.com> wrote:
We have multiple tests that call Toolkit.getDefaultToolkit() for exactly this
purpose (to initialize the native toolkit), so I don't really see a problem
here.
Does an app from the bug report start to work correctly with this fix?
When apps and/or AWT/Swing libs call the Toolkit.getProperties() to get the
values? Does this usually happen after the toolkit is initialized in the real
world apps?
--
best regards,
Anthony
On 09/13/2013 04:06 PM, Leonid Romanov wrote:
I've implemented the approach you suggested, but it didn't pass the regression test I've
written for this bug. Even though the test isn't run in headless mode, it doesn't trigger
LWCToolkit initialization (you have to insert "Toolkit.getDefaultToolkit()"
line to force it). So, I'm unsure whether this approach is a right one. What do you guys
think?
On Sep 12, 2013, at 4:23 PM, Anthony Petrov <anthony.pet...@oracle.com> wrote:
Hi Leonid,
Why not put the OS X specific logic into the LWCToolkit and set the
Toolkit.platformResources field via the AWTAccessor? This would help avoid
checking for a specific platform in our shared code.
--
best regards,
Anthony
On 09/12/13 15:30, Leonid Romanov wrote:
Yes, I'll use OSInfo instead. As for property file name, could you
suggest a better one?
On 12.09.2013, at 15:26, Sergey Bylokhov <sergey.bylok...@oracle.com
<mailto:sergey.bylok...@oracle.com>> wrote:
Hi, Leonid.
i think that you can try to reuse OSInfo class.
Also I doubt about property file name.
On 12.09.2013 9:09, Leonid Romanov wrote:
Hi guys,
Could you review it, please?
-------- Original Message --------
Subject: [8] Review request for 7129133: [macosx] Accelerators are
displayed as Meta instead of the Command symbol
Date: Wed, 11 Sep 2013 13:24:32 +0400
From: Leonid Romanov <leonid.roma...@oracle.com>
To: awt-dev <awt-dev@openjdk.java.net>
CC: macosx-port-...@openjdk.java.net
<macosx-port-...@openjdk.java.net>, build-dev
<build-...@openjdk.java.net>
Hello,
Please review a fix for 7129133: [macosx] Accelerators are displayed as Meta
instead of the Command symbol.
bug:http://bugs.sun.com/view_bug.do?bug_id=7129133
webrev:http://cr.openjdk.java.net/~leonidr/7129133/webrev.00/
Thanks,
Leonid.
--
Best regards, Sergey.
--
Best regards, Sergey.
--
Best regards, Sergey.
