Discrepancies

We will present and discuss the findings. We gathered 862 test cases from the standard ex-amples and through our own development. In the average around 7.8% of the test cases failed. Relatively seen the most test cases failed in the arithmetic theory with 14.9%. The fewest test cases failed in the structure theory with 3.3%. The results are very promising, since they show that we would not have a very long path ahead for full ISO compliance.

Picture 1: Number of Discrepancies

There were no test case failures which we could not explain. We have analysed the failures and come up with a set of findings. For each finding we have compared the Jekejeke Prolog behaviour with the behaviour mandated by the ISO standard. We have then classified the Jekejeke Prolog behaviour into the categories enhancement or limitation. In summary the following findings could be collected from our testing:

Table 1: Identified Findings
Theory Title Classification
Control Predicate Visibility
Enhancement
Error Message
Limitation
Clean-up Safety
Enhancement
Simplify Glitch
Limitation
Consult Predicate Sealing
Enhancement
Arithmetic Narrower Arithmetic Limitation
Broader Arithmetic Enhancement
Structure Array Access Enhancement
Stream
Stream Property Enhancement

The findings that were classified a limitation need further work by us. They will be probably fixed in an upcoming release of Jekejeke Prolog. The findings that were classified an en-hancement will only be worked on, when we have introduce an ISO compatibility flag. This flag will then either allow the Jekejeke Prolog specific behaviour or it will revert to the ISO compliant behaviour. The ISO compliant behaviour will also need further work by us.

We will present our findings grouped according to our packages:

Kommentare