OK, thanks.
Unfortunately I don't think the "static int run()" technique can be applied to
rmiregistry and rmid, since a significant part of testing them -- apart from
their usage messages :-) -- involves operating within a fairly strict security
policy. But it's good to keep this in mind.
s'marks
On 2/20/14 6:36 PM, Mandy Chung wrote:
Thumbs up.
FYI - jdeps follows the langtools convention to define a static int
run(String... args) method that doesn't exit but instead returns the error exit
code (whereas the main method will exit with an error code) so that the test can
test the tool with different options in the same VM and avoid launching another
VM. In any case, I agree these 2 tests are not as useful.
Mandy
On 2/20/2014 6:02 PM, Stuart Marks wrote:
Hi all,
There are a couple of RMI tests that do nothing except check the usage message
emitted by the 'rmiregistry' and 'rmid' commands when given invalid
command-line options. Since two bugs have occurred in these tests, I have
determined that the overhead of maintaining these tests outweighs their
usefulness, and I'm therefore removing them.
Bug:
https://bugs.openjdk.java.net/browse/jdk-8027810
Webrev:
http://cr.openjdk.java.net/~smarks/reviews/8027810/webrev.0/
Thanks,
s'marks
