CLEAR GRAPH not necessarily cleared for all SPARQL queries

[fgiasson]

@HughWilliams, @iv-an-ru,

I am having a really weird issue with my Virtuoso 7 instance. It is using the latest Dev branch code, but I am wondering if the index is not somehow broken.

I did import a dataset using DB.DBA.TTLP_MT.

Then I cleared it using the following SPARQL Update query:

sparql clear graph <http://localhost/datasets/test/>;

Then I ran that query:

sparql
select count(?s) as ?nb
from <http://localhost/datasets/test/>
where
{
  ?s a ?type.
};

I was expecting to get 0, but I got 16 (the number doesn't matter, what matters is that it is no 0 :) )

Then I tested this other query:

select count(?s) as ?nb
from <http://localhost/datasets/test/>
where
{
  ?s ?p ?type.
}

And then I got 0.

So, if I was using a or not was impacting the query to return results or not (even the first thing I did is to clear the graph!).

This is really weird and totally unexpected and this is why I am wondering if this is not due to a corrupted index or something similar. Is there some procedure I could run to detect/fix issues with an index? Can you reproduce this issue?

[kidehen]

On 1/28/15 2:39 PM, Frederick Giasson wrote:

@HughWilliams https://github.com/HughWilliams, @iv-an-ru
https://github.com/iv-an-ru,

I am having a really weird issue with my Virtuoso 7 instance. It is
using the latest Dev branch code, but I am wondering if the index is
not somehow broken.

I did import a dataset using |DB.DBA.TTLP_MT|.

Then I cleared it using the following SPARQL Update query:

|sparql clear graph http://localhost/datasets/test/;
|

Then I ran that query:

|sparql
select count(?s) as ?nb
from http://now.winnipeg.ca/datasets/zoning-parcels/
where
{
?s a ?type.
};
|

I was expecting to get |0|, but I got |16| (the number doesn't matter,
what matters is that it is no |0| :) )

Then I tested this other query:

|select count(?s) as ?nb
from http://now.winnipeg.ca/datasets/zoning-parcels/
where
{
?s ?p ?type.
}
|

And then I got |0|.

So, if I was using |a| or not was impacting the query to return
results or not (even the first thing I did is to clear the graph!).

This is really weird and totally unexpected and this is why I am
wondering if this is not due to a corrupted index or something
similar. Is there some procedure I could run to detect/fix issues with
an index? Can you reproduce this issue?

What happens when you run:

DROP SILENT GRAPH

Regards,

Kingsley Idehen
Founder & CEO
OpenLink Software
Company Web: http://www.openlinksw.com
Personal Weblog 1: http://kidehen.blogspot.com
Personal Weblog 2: http://www.openlinksw.com/blog/~kidehen
Twitter Profile: https://twitter.com/kidehen
Google+ Profile: https://plus.google.com/+KingsleyIdehen/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen
Personal WebID: http://kingsley.idehen.net/dataspace/person/kidehen#this

[fgiasson]

Hi @kidehen,

I tried that too, but I am getting that error:

SQLState: 22023 Message: SPARUL_DROP() failed: graph <http://localhost/datasets/test/> has not been explicitly created before

Is it necessary to create it in a particular manner?

[kidehen]

On 1/28/15 3:35 PM, Frederick Giasson wrote:

Hi @kidehen https://github.com/kidehen,

I tried that too, but I am getting that error:

|SQLState: 22023 Message: SPARUL_DROP() failed: graph http://localhost/datasets/test/ has not been explicitly created before
|

Is it necessary to create it in a particular manner?

No, SILENT should simply stop the aforementioned error occurring i.e.,
don't require that said named graph was explicitly created using CREATE
GRAPH .

Regards,

Kingsley Idehen
Founder & CEO
OpenLink Software
Company Web: http://www.openlinksw.com
Personal Weblog 1: http://kidehen.blogspot.com
Personal Weblog 2: http://www.openlinksw.com/blog/~kidehen
Twitter Profile: https://twitter.com/kidehen
Google+ Profile: https://plus.google.com/+KingsleyIdehen/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen
Personal WebID: http://kingsley.idehen.net/dataspace/person/kidehen#this

[fgiasson]

@kidehen,

thanks for clarifying for the silent clause. However, exactly the same behavior happens.

Could it be possible that it is because the database comes from a V6 installation (even if I think they are supposedly compatible)?

[fgiasson]

Another thing I just noticed, the dataset I just dropped still appears in the resultset of that query:

select distinct ?g
where
{
    graph ?g {?s ?p ?o}
}

[fgiasson]

@kidehen, @HughWilliams,

I exported the datasets, created a new empty V7.1 database, re-imported the datasets in it, and now I cannot reproduce this issue. I think the DB has been corrupted in some ways which was leading to these weird behaviors.

Thanks!