We have a PIT (pre-integration-testing process) but if that these clashes are presented only as an FYI and not an actual failure then SQE might not notice it. It may even get swallowed and not forwarded for human inspection by some layered tool on top. I don't actually know .. I am just speculating.
-phil. On 11/23/16, 6:50 PM, Martin Buchholz wrote:
On Wed, Nov 23, 2016 at 1:34 PM, Phil Race <[email protected] <mailto:[email protected]>> wrote:Hi, So your real complaint isn't the failure itself, but that jtreg bothers to check directories you aren't even running tests from, and that is a time tax whether such clashes exist or not ? Seems a reasonable point ... if I'm running a single Image I/O test jtreg still finds the java/awt clash and that can't be "free". That was _one_ of my complaints!I'm also unhappy that jdk9/dev got poisoned by jdk9/client. Having subforests is supposed to prevent that sort of breakage via proper release engineering. I'm _not_ unhappy about the mistake itself - it's a classic software engineering trap we all fall into eventually if we do software long enough.
