Re: test cases on wiki

On 21 May 2008, at 15:06, Ian Horrocks wrote:

> On 19 May 2008, at 16:34, Bijan Parsia wrote:
[snip]
>> Writing test cases really really sucks, by the by.
>
> Can you say a little more about what the problems are and how you  
> think they might be addressed?

Sure. There's lots of axis (expressivity, kind of test, granularity  
etc.) of metadata and a lot of metadata that isn't obvious outside  
some arbitrary decisions (e.g., naming the test case, providing a  
description). To convert my table of different bnode semantics and  
their effects would require a pretty large effort and up to twelve  
wiki pages (and it's weird, can I use the same test case to cover the  
cases where both results are identical under either semantics? should  
I only produce a test case for the situations where they differ?)

Some of it is the wiki; some of it is the difficulty of putting  
together a nice framework; and some of it is that writing test cases  
is inherently hard.

>> I strongly request of the chairs that we do something to move this  
>> forward. Test cases written in a crunch at the end are going to be  
>> painful to do and hard to validate. I suggest that we actually put  
>> some people in charge of doing this.
>
> We are on to it.

Great!

>> I am not one of those people!
>
> Agreed.

Even greater!

Cheers,
Bijan.

Received on Wednesday, 21 May 2008 14:24:48 UTC