Thanks for doing this. Looks reasonable to me.
(It's also not clear to me that a regression test is worthwhile for testing
message output like this. In the RMI tests, we used to have tests that tested
the usage messages emitted by the launchers. The tests had race conditions that
led to intermittent failures! The "fix" was to remove them. But I'll leave it up
to you folks as to whether such regression tests are valuable for your area.)
s'marks
On 4/28/16 10:22 AM, Sergey Bylokhov wrote:
Hello.
Please review the small fix for jdk9.
The message about deprecation of appletviewer now will be printed to err. the
flush() was added, so the message will not be mixed with some normal output.
The reason of the change was provided by Stuart:
"Standard output is used for the application's data output. If the deprecation
warning were sent to standard output, it would be intermixed with any output
from the application. This might cause errors downstream, when some other
program tries to process this application's output, and unexpectedly encounters
the deprecation warning. Standard error is used not only for actual errors, but
for other diagnostics, meta-information, etc., including warning messages.
"
Bug: https://bugs.openjdk.java.net/browse/JDK-8154493
Webrev can be found at: http://cr.openjdk.java.net/~serb/8154493/webrev.00
