Re: Turtle ResPec version

On 04-01-14 14:55, Eric Prud'hommeaux wrote:
> * Guus Schreiber <guus.schreiber@vu.nl> [2014-01-04 11:43+0100]
>>
>>
>> On 03-01-14 23:26, Eric Prud'hommeaux wrote:
>>> * Guus Schreiber <guus.schreiber@vu.nl> [2014-01-03 18:28+0100]
>>>> ...
>>>> 2. Broken fragments for tests
>>>>
>>>> These links are not valid anymore in the new test setup. Remove?
>>>
>>> Hmm, tricky to deal with the legacy of my mistake from years ago. The
>>> CR doc linked to them and inappropriately referred to them as
>>> tests. Because they weren't the official test suite, they served only
>>> as examples. The directory is there in perpetuity and probably the
>>> most honest thing to do is call them examples in the text à la:
>>>
>>> s/(test: <a href="tests/#base1">base1</a> <a href="tests/#base2">base2</a>)
>>>   /(example: <a href="tests/#base1">base1</a> <a href="tests/#base2">base2</a>)
>>>   /
>>>
>>> and change the http://www.w3.org/TR/turtle/tests/ document to say:
>>> [[
>>> <h1>Turtle Examples</h1>
>>>
>>> The following pairs of Turtle and N-Triples documents serve as
>>> auxiliary examples to the Turtle specification. (Despite the directory
>>> name, these are not official Turtle tests.)
>>>
>>> <h2>Example Coverage</h2>
>>>
>>> The following table associates example labels with the features those
>>> examples are supposed to cover.
>>> ]]
>>>
>>> Shall I?
>>
>> This looks like a reasonable solution, assuming it is OK to touch
>> this document in TR space.
>
> Any future pulication would overwrite the directory for
> /TR/turtle/tests so the act of publishing grants us that
> permission. Updating them will obviate Gavin's action 281: "Provide
> sandro text for http://www.w3.org/TR/turtle/tests/ to NOT say it's the
> test suite"
>    <http://www.w3.org/2011/rdf-wg/track/actions/281>
>
> I'm just trying to figure out the mechanics. Usually when I publish, I
> copy all the appropriate files around. We don't have a tests directory
> in the Mercurial repo so I can either revert the earlier deletion (or
> however one does that in Mercurial), or, reply to this thread with a
> copy of the new tests/Overview.html and drop it into place when we're
> ready to publish. My preference is the latter as it involves less
> negotiation with mercurial.

I've reread the text in Sec.  7.1. On second thought, I don't see what 
these tests add.
The textual explanation is crystal clear. Real tests for this should be 
in the test suite. I think it would be much easier and cleaner to just 
drop these tests from the main document.

Guus

Received on Saturday, 4 January 2014 21:41:12 UTC