Re: UUID per binding

[Andy Seaborne]

On 26/10/2019 21:24, Martynas Jusevičius wrote:

OK. So I guess I’m out of luck.

Is this under-specified in SPARQL 1.1? Something to be addressed in 1.2?

No and no. As stated the twitter thread you quoted.
https://twitter.com/AndySeaborne/status/876157902125387776

SPARQL 1.2 isn't the place to dump implementation bugs.

    Andy

On Sat, 26 Oct 2019 at 21.58, Andy Seaborne <a...@apache.org> wrote:

JENA-880

On 26/10/2019 20:43, Martynas Jusevičius wrote:

Hi,

I have this simplified query

PREFIX owl: <http://www.w3.org/2002/07/owl#>

CONSTRUCT
{
      ?s <smth> ?ab.
}
{
    BIND (STRUUID() AS ?a)
    ?s a owl:Class .
    OPTIONAL {
      ?s ?p ?o .
      BIND (STRUUID() AS ?b)
    }
    OPTIONAL {
      ?x ?y ?z .
      BIND (STRUUID() AS ?c)
    }
    BIND (CONCAT(?a, "/", ?b) AS ?ab)
    BIND (CONCAT(?a, "/", ?c) AS ?ac)
}

What I'm getting is multiple ?ab UUIDs for the same ?s binding:

skos:Collection  <http://server/unset-base/smth>

"520a2768-dd5e-4a5f-b2a8-09328ad89b3f/8d946969-92f2-4239-bad0-b0c92c606b9e"

,

"520a2768-dd5e-4a5f-b2a8-09328ad89b3f/ce7863e0-33b6-47bf-bae3-e85b3e350636"

,

"520a2768-dd5e-4a5f-b2a8-09328ad89b3f/b06ece07-cdbb-4196-ae34-f7676f739b55"

,

"520a2768-dd5e-4a5f-b2a8-09328ad89b3f/2953ef49-775c-4f52-bdd0-f5b8f0cf82d6"

,

"520a2768-dd5e-4a5f-b2a8-09328ad89b3f/0eb8c370-a672-4476-b1fe-f8e9acb93901"

,

"520a2768-dd5e-4a5f-b2a8-09328ad89b3f/630bc8cb-30f0-4d87-93d1-0f24c432bc5f"

,

"520a2768-dd5e-4a5f-b2a8-09328ad89b3f/f318a7f4-8510-4e42-ad60-18a711e6af0e"

,

"520a2768-dd5e-4a5f-b2a8-09328ad89b3f/1d78b7ba-c8f7-42fb-8c72-a2f85732c8a0"

,

"520a2768-dd5e-4a5f-b2a8-09328ad89b3f/5b606245-e5c1-4aa7-a98c-36bada0c21ef"

,

"520a2768-dd5e-4a5f-b2a8-09328ad89b3f/01148de7-8ebb-4a6f-843f-8664a27e2eee"

.

I would expect to get only one UUID per ?s. Can the query be
reformulated to achieve this?
I cannot move ?ab and ?ac inside the patterns next to STRUUID()
because ?a is undefined there.

Doing SELECT (COUNT(DISTINCT ?ac) AS ?count) returns 11655 results
while it only returns 315 on Dydra. I think the difference is related
to this issue:
https://github.com/w3c/sparql-12/issues/102

I ran it on sparql.org over the skos.owl:

https://raw.githubusercontent.com/usc-isi-i2/karma-step-by-step/master/preloaded-ontologies/skos.owl

Martynas