On 09/15/2014 02:42 PM, Alexander Scherbatiy wrote:
test/lib/testlibrary/jdk/testlibrary/OSInfo.java
32 import static jdk.testlibrary.OSInfo.OSType.*;
Is this import necessary?
Yes: many tests use enum without qualification -- it's
short and presumably elegant. Without this import
usage like OSInfo.OSType.WINDOWS is necessary (see an
example of changed test in this webrev).
115 public static PrivilegedAction<OSType> getOSTypeAction() {
116 return osTypeAction;
117 }
Is this method used in the tests?
Well, no. Almost never. This method of sun.awt.OSInfo is in use
in Swing classes.
I'm retaining it because (1) no harm in it and (2) this way,
it would be much simpler to maintain a duplicate of
sun.awt.OSInfo in case of a change.
-yan
Thanks,
Alexandr.
On 9/15/2014 11:35 AM, Yuri Nesterenko wrote:
Dear friends, one more weekly reminder!
Without this (or similar) fix applied, we cannot start changes of
~60 regtests, and time is short.
Cheers,
-yan
On 09/08/2014 01:09 PM, Yuri Nesterenko wrote:
Weekly reminder!
Cheers,
-yan
On 09/01/2014 12:44 PM, Yuri Nesterenko wrote:
Colleagues,
please review this minimal change to fix
https://bugs.openjdk.java.net/browse/JDK-8056991
http://cr.openjdk.java.net/~yan/8056991/webrev.00
In the webrev there is an example of a test refactored.
We need to clean up regression tests from internal
dependencies. One of them, dependency on sun.awt.OSInfo.java, a
standard (however internal) tool to provide a version
of current OS.
Here, I'm just copying the file to a test helper directory.
(1) no swing library class depending on OSInfo will be affected
(2) no public API change occurs
A person updating sun.awt.OSInfo in future should, however,
duplicate changes in this test copy as well which is an ugly
compromise.
Thanks,
-yan
