possible bug in SPARQL syntax test suite

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

possible bug in SPARQL syntax test suite

Jeen Broekstra-5
In running the syntax tests over the latest update of the SPARQL test
suite, I encountered the following positive test, which fails for me:

http://www.w3.org/2009/sparql/docs/tests/data-sparql11/syntax-query/qname-escape-01.rq

The query is the following:

PREFIX og: <http://ogp.me/ns#>
SELECT * WHERE {
        ?page og:audio\:title ?title
}

As far as I can tell, this test case is out of date: colons in prefixed
localnames are no longer to be escaped, and grammar production
PN_LOCAL_ESC (see http://www.w3.org/TR/sparql11-query/#rPN_LOCAL_ESC)
does no longer contain the ':' character as a legal value for escaping.

There is a negative syntax test which covers exactly the same issue:

http://www.w3.org/2009/sparql/docs/tests/data-sparql11/syntax-query/syn-bad-pname-06.rq

Query:

PREFIX : <http://example/>
SELECT *
{
  :a :b :c\:z .
}

Clearly, these two tests are contradicting each other, and I suspect
that the first one should be removed.

Regards,

Jeen

Reply | Threaded
Open this post in threaded view
|

Re: possible bug in SPARQL syntax test suite

Andy Seaborne-3
Correct - this is :test_52 in the manifest and it's commented out of the
list of tests in the manifest:

The manifest has:

[[
# :test_52 # obsoleted by decision to allow colons in prefixname local
parts, but not allow them to be escaped
]]


Do you drive your test harness off the manifest or directly from the
files in the directory?

(Personally, I wanted to remove tests that weren't active)

        Andy


On 18/12/12 01:36, Jeen Broekstra wrote:

> In running the syntax tests over the latest update of the SPARQL test
> suite, I encountered the following positive test, which fails for me:
>
> http://www.w3.org/2009/sparql/docs/tests/data-sparql11/syntax-query/qname-escape-01.rq
>
> The query is the following:
>
> PREFIX og: <http://ogp.me/ns#>
> SELECT * WHERE {
> ?page og:audio\:title ?title
> }
>
> As far as I can tell, this test case is out of date: colons in prefixed
> localnames are no longer to be escaped, and grammar production
> PN_LOCAL_ESC (see http://www.w3.org/TR/sparql11-query/#rPN_LOCAL_ESC)
> does no longer contain the ':' character as a legal value for escaping.
>
> There is a negative syntax test which covers exactly the same issue:
>
> http://www.w3.org/2009/sparql/docs/tests/data-sparql11/syntax-query/syn-bad-pname-06.rq
>
> Query:
>
> PREFIX : <http://example/>
> SELECT *
> {
>    :a :b :c\:z .
> }
>
> Clearly, these two tests are contradicting each other, and I suspect
> that the first one should be removed.
>
> Regards,
>
> Jeen
>

Reply | Threaded
Open this post in threaded view
|

Re: possible bug in SPARQL syntax test suite

Jeen Broekstra-5
On 18/12/12 22:46, Andy Seaborne wrote:

> Correct - this is :test_52 in the manifest and it's commented out of the
> list of tests in the manifest:
>
> The manifest has:
>
> [[
> # :test_52    # obsoleted by decision to allow colons in prefixname
> local parts, but not allow them to be escaped
> ]]
>
>
> Do you drive your test harness off the manifest or directly from the
> files in the directory?

Ah! Thanks for the tip.

I do use the manifest, but I don't use the list of tests to retrieve
individual cases, I just query for instances of PositiveSyntaxTest (or
NegativeSyntaxTest).

Still, I can relatively easily adapt that to check that the test occurs
in the list.

> (Personally, I wanted to remove tests that weren't active)

That seems more sensible, but now that I know the modus operandi I can
work with this.

Thanks,

Jeen


Reply | Threaded
Open this post in threaded view
|

Re: possible bug in SPARQL syntax test suite

Andy Seaborne-3
FYI:

There's a proposal to carefully tidy up some more so that unlinked,
unapproved test don't get in the way:

http://lists.w3.org/Archives/Public/public-rdf-dawg/2012OctDec/0228.html

which has a patch attached.

        Andy

On 18/12/12 22:53, Jeen Broekstra wrote:

> On 18/12/12 22:46, Andy Seaborne wrote:
>> Correct - this is :test_52 in the manifest and it's commented out of the
>> list of tests in the manifest:
>>
>> The manifest has:
>>
>> [[
>> # :test_52    # obsoleted by decision to allow colons in prefixname
>> local parts, but not allow them to be escaped
>> ]]
>>
>>
>> Do you drive your test harness off the manifest or directly from the
>> files in the directory?
>
> Ah! Thanks for the tip.
>
> I do use the manifest, but I don't use the list of tests to retrieve
> individual cases, I just query for instances of PositiveSyntaxTest (or
> NegativeSyntaxTest).
>
> Still, I can relatively easily adapt that to check that the test occurs
> in the list.
>
>> (Personally, I wanted to remove tests that weren't active)
>
> That seems more sensible, but now that I know the modus operandi I can
> work with this.
>
> Thanks,
>
> Jeen
>
>