re DERBY-88

[Myrna van Lunteren]

Hi,

 

I happened to notice that indeed, the problem as described is still occurring.

 

If I understand the problem, we don't actually want 'accurate' rounding when the percentage of failure gets small...?

 

We don't want to get the impression that all tests passed (100%) in combination with seeing a failure.

I am wondering if a forced adjustment of the percentage would be acceptable?

 

I am thinking of the following adjustment to ../functionTests/harness/GenerateReport.java:

-----------------------------------------------------

@@ -187,7 +187,13 @@
                NumRun = NumPass+NumFail;
                NumSkip = CountLines (skipFileName);
                PercentPass = (int)Math.ceil(100* ((double)NumPass/(double)NumRun));
+               // use brute force to not get 100% pass if there's even 1 failure.

+               if ((PercentPass == 100) && (NumFail != 0))
+                       PercentPass = 99;
                PercentFail = (int)Math.floor(100* ((double)NumFail/(double)NumRun));
+              // use brute force to not get 0% fail if there's even 1 failure.

+               if ((PercentFail == 0) && (NumFail != 0))
+                       PercentFail = 1;
        }

------------------------------------------------------

 

If something like this would be ok, I can make a patch for it...