W3C home > Mailing lists > Public > public-rdf-dawg-comments@w3.org > November 2011

Re: Tests assume distinct output rows are a bad idea

From: Lee Feigenbaum <lee@thefigtrees.net>
Date: Tue, 01 Nov 2011 11:00:39 -0400
Message-ID: <4EB00997.9070701@thefigtrees.net>
To: Dave Beckett <dave@dajobe.org>
CC: public-rdf-dawg-comments@w3.org
Hi Dave,

Could you please supply us with another example test that you have 
concerns about?

many thanks,
Lee
On Behalf of the SPARQL WG

On 11/1/2011 10:58 AM, Dave Beckett wrote:
> Nope.
>
> My comment was about tests using SHA224 as an example not about SHA224.
> Please confirm that the tests no longer rely on distinctness of the output
> triples.
>
> Dave
>
>
> On 11/1/11 7:25 AM, Lee Feigenbaum wrote:
>> Hi Dave,
>>
>> The Working Group has recently resolved to remove the SHA224 function from
>> SPARQL 1.1 Query. As a byproduct of this, the test in question below has
>> also been removed from the test suite, so I think that the below issue is no
>> longer relevant.
>>
>> If you could reply to the list to acknowledge that this addresses your
>> comment, we'd appreciate it.
>>
>> best,
>> Lee
>> On behalf of the SPARQL WG
>>
>> On 10/9/2011 1:48 PM, Dave Beckett wrote:
>>> A lot of the functions/* tests such as sha224-02 got unicode added
>>> recently which assumes the output result rows are distinct.  This broke
>>> many of my rasqal tests.
>>>
>>> This is a bad idea for two reasons
>>> 1) you should not make this assumption
>>> 2) it does not distinguish the state of the two tests in the file - the \u
>>> form or the unicode form.  If something broke, you couldn't tell since
>>> only 1 result is expected
>>>
>>> so instead of:
>>> :s8 :str "\u98DF", "食" .
>>>
>>> use
>>> :s8 :str "\u98DF" .
>>> :s9 :str  "食" .
>>>
>>> and adjust the output results to suit
>>>
>>> Thanks
>>>
>>> Dave
>>>
>>>
>>>
>>
>
>
Received on Tuesday, 1 November 2011 16:09:23 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 1 November 2011 16:09:24 GMT