Hello,
There is a high barrier to consistent use of new jtreg keywords. FWIW, I
would prefer expanding the scope of @key randomness to defining a new
@key nondeterminstic.
Thanks,
-Joe
On 11/18/2015 12:41 AM, Paul Sandoz wrote:
On 18 Nov 2015, at 03:46, Martin Buchholz <marti...@google.com
<mailto:marti...@google.com>> wrote:
On Tue, Nov 17, 2015 at 2:26 AM, Paul Sandoz <paul.san...@oracle.com
<mailto:paul.san...@oracle.com>> wrote:
For the jtregTest updates, if you have any excess energy
remaining, you might want to consider adding the jtreg randomness
tag (i cannot recall its exact name) to the test metadata.
Most of the jtreg tests deliberately include non-determinism from use
of multiple threads, and there is no @key nondeterminstic tag, and
adding randomness tags to the tests that additionally invoke a prng
doesn't really seem useful. Flakes R Us!
I am sure Joe, CC’ed will have an opinion on this. IIUC correct the
existing randomness tag actually serves a simialr purpose as your
proposed nondeterminstic tag.
I support adding @key nondeterministic, and allowing that a tag to
apply to an entire tree (java.util.concurrent). Do we have any tests
dependent on the phase of the moon?
Dunno :-) perhaps every one in a while a machine running tests is
effected by neutrinos and causes a one off error?
Paul.
