Re: Best practice for RDF test manifests

Note that they general way a JSON-LD version of the manifest is by framing, and this will not include tests that aren’t in entries. I would go for (3) as being least disruptive, but i agree with Andy that there is a range of changes that could be made depending on how entrenched the test is.

I wouldn’t change a former positive test to negative, but perhaps would create a new negative test, if warranted. 

Gregg Kellogg

Sent from my iPad

> On Mar 24, 2021, at 4:34 AM, Pierre-Antoine Champin <pierre-antoine.champin@ercim.eu> wrote:
> 
> Thanks Andy,
> 
>> On 24/03/2021 12:02, Andy Seaborne wrote:
>> How about this decision tree?
>> AKA "It depends"
>> 
>> 
>> Has it become a negative test?
>>   Mark as negative test. (4)
> 
> I am not comfortable with changing a test from positive to negative. This means that an EARL report asserting a "PASS" would mean the exact opposite depending on the version of the test-suite that you are considering...
> 
> I have one such test -- I decided instead to withdraw the positive test and make a new negative test.
> 
>> 
>> Was the suite ever "published" with this test in it?
>>   If not, remove. (1)
>> 
>> Are there any reports that mention the test?
>>   If not, remove. (1)
>> 
>> Else
>> (2)+(3).
> 
> makes sense :)
> 
>> 
>> Removing from the entries list is quite important.
>> 
>>     Andy
>> 
>>> On 24/03/2021 10:44, Pierre-Antoine Champin wrote:
>>> Hi RDF testers,
>>> 
>>> I am currently updating the test-suite of RDF-star [1], and I am not sure  about the best way to go to remove a test from the suite.
>>> 
>>> Here are the options that I can think of:
>>> 
>>> 1) Removing the test completely from the manifest file
>>> PROS: keeps the manifest clean (and I still have archives in git).
>>> CONS: old EARL reports will point to something that does not exist anymore. Cool URIs don't change (or disappear).
>>> 
>>> 2) Keep the test in the manifest.ttl file, but remove it from the entry list.
>>> PROS: will be easily ignored by people running the test suite, but still findable from old EARL reports.
>>> CONS: clutters the file.
>>> 
>>> 3) Mark the test as Withdrawn (one of the statuses defined by [2]).
>>> PROS: semantically clean and "sufficient".
>>> CONS: clutters the file and the manifest entry list; forces users to look at the status when iterating over the entries.
>>> 
>>> For the moment, I went for 2+3, to be on the safe side. Would you agree with this approach? Is there another option that I didn't think of?
>>> 
>>> Thanks in advance.
>>> 
>>>    pa
>>> 
>>> 
>>> [1] https://w3c.github.io/rdf-star/
>>> 
>>> [2] http://www.w3.org/2001/sw/DataAccess/tests/test-dawg#
>>> 
>>> 
>>> 
>> 
>

Received on Wednesday, 24 March 2021 14:50:37 UTC