Sounds sensible, I will try and take a look Monday Rob ________________________________________ From: Andy Seaborne [[email protected]] Sent: 31 August 2013 04:28 To: [email protected] Subject: Re: jena-jdbc : JDJK6->JDK7-ism?
Some progress. I have a version of the test suite that runs, without inserted pauses, in Java6 and Java7. It's not perfect - you can't always run the test suite immediately after it's just run. It may need a couple of minutes between runs for all the sockets in TIME_WAIT timeout (the JVM has exited - this is kernel resources). Sometimes, just the later tests (mem, TDB) is long enough and test can be run back-to-back. The change is that there are now operations in HttpOp to get/set the default HttpClient. Hackery in AbstractRemoteEndpointResultSetTests sets a single default caching HttpClient used for all non-auth tests (HttpOp does not use the default HttpClient for auth requests so as to get containment). I was confused by the fact that TS_JdbcDriverRemote isn't called by maven. I have created https://issues.apache.org/jira/browse/JENA-527 and attached a patch. The entire "mvn clean test" of jena-jdbc takes 5m15s. The longest part is the TDB tests (4m30s). This needs reviewing in case I have compromised testing, and also testing on other machines. Rob - is this an acceptable/useful thing to do for the jena-jdbc tests? Andy
