Hi Semyon,
(see also my answer in
https://bugs.openjdk.java.net/browse/JDK-8152693 )
Unfortunately, this changeset is already pushed.
Generally, some of the tests is not hard to adapt for non-modules
execution mode, and let's try that sometimes, however
(1) some of them already run well with jdk8 (bug8133039.java,
PointerInfoCrashTest.java);
(2) there's no way back from Jigsaw; with time, that conditional
execution would introduce unnecessary complexity
(WrappedToolkitTest.sh would be two times bigger, why not use
jdk8 version?).
(3) Some of the tests is not easy to make two-way even now:
for instance, every test requiring Xpatch.
Thank you! In future I will have in mind that getLocation()
issue.
-yan
On 03/29/2016 10:49 AM, Semyon Sadetsky wrote:
Hi Yuri,
Is it possible to make those tests to run on both Jigsaw and non-Jigsaw
JDKs?
Some tests may be unstable on Ubuntu because of
https://bugs.openjdk.java.net/browse/JDK-8036915.
To make stability better I propose to replace all top windows
getLocation() by getLocationOnScreen() and for rectangles got by
getBounds() to call setLocation(window.getLocationOnScreen()).
getLocationOnScreen() queries the position from WM directly and is more
reliable.
--Semyon
On 3/24/2016 6:12 PM, Yuri Nesterenko wrote:
Sergey, I'm sorry,
filing JDK-8152693, I found an error in that version.
A new one is
http://cr.openjdk.java.net/~yan/8150535/webrev.02
The spawned processes were provided with addExports.
Shame on me!
-yan
On 03/24/2016 05:34 PM, Sergey Bylokhov wrote:
cc swing-dev
On 24.03.16 17:17, Yuri Nesterenko wrote:
Hi,
please review this test-only fix, a leftover from Jigsaw M3
integration.
Some 7 tests fixed but note that not all of them always pass even
fixed.
Webrev:
http://cr.openjdk.java.net/~yan/8150535/webrev.01
I need to file a bug about the fixed here manual test
java/awt/xembed/server/TestXEmbedServerJava
Thank you!
-yan
