Re: Additional tests for cyclic structures

On Mon, 29 Sep 2003, Jeremy Carroll wrote:

> I earlier reported that my code did not pass the test for cyclic structures:
> I fixed my code today. As I was half way through fixing the code, it passed
> that test and did not fail any others (well that part of the code).
> A colleague argues that you should only write code to pass tests - so I wrote
> a few more tests to fail before writing the half of the cycle detection fix
> to my code.
> These are:
>   I5.26-008
>     A cycle of unnamed individuals
>   I5.5-005
>    Well formed list comprehension
>   I5.5-006
>    A cylic list
>   I5.5-007
>    A list with itself inside an element.
> The last three also investigate further the list comprehension rules.
> (My code failed those tests and so I fixed it).
> I imagine parsers such as Sean's and Peter's code cope with these no problem.

You have a somewhat over-active imagination, Jeremy :-). Well, it works
now, anyway. I have also added some minor fixes and regenerated test
results based on the latest test manifest.

In the spirit of transparency, I have added a simple test harness to our
servlet interface [1]. If you provide the servlet with the URI of a test
manifest, it will report on whether or not it thinks it passes the syntax
validation of the test (along with some diagnostic information as to why
it has/hasn't).

If anyone tries this and discovers any unexpected failures (or disagrees
with the explanations), do please let me know...




Sean Bechhofer

Received on Thursday, 2 October 2003 10:14:52 UTC