Re: TEST

On May 26, Jeremy Carroll writes:
> 
> 
> Ian:
> [[
> "A complete OWL Lite consistency checker should not return Unknown on
> the OWL Lite consistency or inconsistency tests."
> 
> Although not ambiguous in the context of the other statements in the
> document, it might be worth emphasising that this is the case
> *regardless* of the datatypes supported/used.
> ]]
> accepted
> 
> "A complete OWL Lite consistency checker should not return Unknown on
> the OWL Lite consistency or inconsistency tests, regardless of the use of
> unsupported datatypes."
> 
> [[
> Regarding the text:
> 
> + If there are any datatypes that are required to be not supported by the test 
> then at least one of them is not supported by the datatype theory of the 
> checker. 
> 
> Is this correct 
> ]]
> 
> Yes with a closed world assumption, no without.
> Since the sentence implicitly refers to a finite set of tests (I believe there 
> is only one test with more than one notSupportedDatatype) we can exhaustively 
> check, which I have done.
> 
> If we decide that the sort of test you describe is desirable then this section 
> will need work: I suggest not doing that work until we have such a test (and 
> if we had one such test only I might be inclined to highlight as an exception 
> as opposed to generalizing the text, rather like imports-002).

This seems to be a vary strange decision. Surely the text should try
to be as general as possible so that tests can be added without having
to check/update the text every time. Moreover, I found this particular
piece of text very confusing as it never occurred to me that it might
refer only to the current set of tests.

Why not change the text to:

+ If there are any datatypes that are required to be not supported by the test,
  then they must not be supported by the datatype theory of the 
  checker (in some cases it may be sufficient for only some of them to
  be not supported).


Ian


> 
> Jeremy
>  
> 
>

Received on Monday, 26 May 2003 16:08:00 UTC