Fix to mend debug logging to allow programmatic level setting.

[whitingjr]

This PR fixes changes introduced in 1205 release.
It adds a new method to PGProperty to return a property value without using the default. Allowing NULL to be returned.
Changed the impl of isPresent method. To actually work.
Added two test cases. To check isPresent and getSetString methods both work.

The fixed isPresent method smokes out an issue in an SSL test case. Now the method is fixed this test fails now.

[davecramer]

@whitingjr not sure why it failed Travis CI ?

[lordnelson]

The tests failed:

[junit] Testsuite: org.postgresql.test.jdbc2.Jdbc2TestSuite
    [junit] Tests run: 431, Failures: 0, Errors: 1, Time elapsed: 137.177 sec
    [junit] 
    [junit] ------------- Standard Output ---------------
    [junit] file:/home/travis/build/pgjdbc/pgjdbc/jars/postgresql-9.4-1206.jdbc4.jar
    [junit] ------------- ---------------- ---------------
    [junit] Testcase: testIsPresentWithParseURLResult(org.postgresql.test.jdbc2.PGPropertyTest):    Caused an ERROR
    [junit] SSL property should be present
    [junit] java.lang.AssertionError: SSL property should be present
    [junit]     at org.postgresql.test.jdbc2.PGPropertyTest.testIsPresentWithParseURLResult(PGPropertyTest.java:212)

[whitingjr]

That testcase failed because the isPresent method was not working correctly beforehand. Then my commit corrected it.
But that caused testIsPresentWithParseURLResult() to start failing.
I've updated the testcase to work as I think it was intended.

[davecramer]

Thanks!!!

Dave Cramer

On 2 December 2015 at 10:44, Jeremy Whiting notifications@github.com
wrote:

That testcase failed because the isPresent method was not working
correctly beforehand. Then my commit corrected it.
But that caused testIsPresentWithParseURLResult() to start failing.
I've updated the testcase to work as I think it was intended.

—
Reply to this email directly or view it on GitHub
#438 (comment).

[davecramer]

Looks like it is still failing

Dave Cramer

On 2 December 2015 at 10:54, Dave Cramer davecramer@gmail.com wrote:

Thanks!!!

Dave Cramer

On 2 December 2015 at 10:44, Jeremy Whiting notifications@github.com
wrote:

That testcase failed because the isPresent method was not working
correctly beforehand. Then my commit corrected it.
But that caused testIsPresentWithParseURLResult() to start failing.
I've updated the testcase to work as I think it was intended.

—
Reply to this email directly or view it on GitHub
#438 (comment).

[whitingjr]

Yes, looks like the CI build has ssl tests enabled. Which means the ssl property is set when the jvm starts.
Do you see any way around this except remove the individual assertion ?

[whitingjr]

I see a way. Use the System.setProperties method to replace all the props.

[davecramer]

seems like you should check for null before checking instanceof ?

[whitingjr]

shouldn't be necessary. instanceof only works on the reference. not the object.
in the situation the reference is null then the instanceof condition returns false.

[vlsi]

Please use spaces, not tabs

[whitingjr]

Yes will replace them with spaces.

[vlsi]

Do you really mean "double check" here?

[whitingjr]

Well spotted. No I hadn't meant to duplicate line 238. Will remove the duplicate.

[davecramer]

We must be able to avoid the NPE in the code, no ?

Dave Cramer

On 2 December 2015 at 11:41, Jeremy Whiting notifications@github.com
wrote:

In org/postgresql/test/jdbc2/PGPropertyTest.java
#438 (comment):

• /**

• \* Check whether the isPresent method really works.
  

• */
  

• public void testPresenceCheck()
• {

•    Properties empty = new Properties();
  

•    Object value = PGProperty.LOG_LEVEL.get(empty);
  

•    assertNotNull(value);
  

•    Assert.assertFalse(PGProperty.LOG_LEVEL.isPresent(empty));
  

• }

• public void testNullValue()
• {

•    Properties empty = new Properties();
  

•    assertNull(PGProperty.LOG_LEVEL.getSetString(empty));
  

•    assertNull(PGProperty.LOG_LEVEL.getSetString(empty));
  

Well spotted. No I hadn't meant to duplicate line 238. Will remove the
duplicate.

—
Reply to this email directly or view it on GitHub
https://github.com/pgjdbc/pgjdbc/pull/438/files#r46438994.

[davecramer]

getString() checks instanceof String before checking for null

Dave Cramer

On 2 December 2015 at 11:13, Jeremy Whiting notifications@github.com
wrote:

I see a way. Use the System.setProperties method to replace all the props.

—
Reply to this email directly or view it on GitHub
#438 (comment).

[vlsi]

Could you please move that to separate test method?

[whitingjr]

@davecramer "getString() checks instanceof String before checking for null"
sorry but you lost me. Which class are you referring to ?

[davecramer]

Looks like this is passing now.

[whitingjr]

Is there anything else considered necessary to make this PR suitable to pull into master ?

[davecramer]

no, it's fine. I was caught up with the maven pull