On 2 Jun 2016, at 14:37, Roger Riggs <roger.ri...@oracle.com> wrote: > > Hi Chris, > > looks fine but a couple of questions/suggestions.
Thanks for looking at this Roger. > CookieManagerTest.java: > > - the new definition of localHostAddr could probably be moved to a class > static and be used instead of the "127.0.0.1" literal (line 97). Agreed. In fact we can do better; where possible the host IP will be used. The test requires an IP literal, for cookie domain matching. Otherwise, the static test string values will all need to be updated. > - Does the use of the IPV4 loopback address work even in the case of > supporting only IPV6? > I'm not sure this is relevant to recent changes to support IPV6 only. Probably not, but while we are moving towards support for IPv6-only environments at runtime, our tests will require a lot of work if they are to be able to be run successfully in such an environment. > UnmodifiableMaps and HandleContentTypeWithAttrs: > > - I assume that the lookup of localhost doesn't trigger the potential > timeouts to be avoided. > (Or 127.0.0.1 would have been used here too). It works fine all systems I’ve tested with. Update in-palce: http://cr.openjdk.java.net/~chegar/8158525/ -Chris. > Roger > > > > On 6/2/2016 9:22 AM, Chris Hegarty wrote: >> The following networking tests have been seen to fail from a timeout, when >> the host >> is not accessible. These tests have nothing whatsoever to do with testing of >> network >> connections, they simply want to test the HTTP client implementation. This >> issue >> proposes to change the tests from using the host's name/address to using the >> loopback, to make them more robust. >> >> test/java/net/CookieHandler/CookieManagerTest.java >> test/java/net/HttpURLConnection/UnmodifiableMaps.java >> test/java/net/URLConnection/HandleContentTypeWithAttrs.java >> >> http://cr.openjdk.java.net/~chegar/8158525/ >> https://bugs.openjdk.java.net/browse/JDK-8158525 >> >> -Chris. >