You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@jena.apache.org by Federico López <fi...@gmail.com> on 2012/06/10 19:36:56 UTC

ARQ Takes to much time to parse a query.

I need to send a query to Fuseki server from Java ans ARQ is taking too
much time parsing and executing a simple query to Fuseki Server. This is my
code:

long before=new java.util.Date().getTime();
        Query query =
QueryFactory.create(OntologyAcces.PREFIXES+"SELECT *{?s ?p ?o} LIMIT
1000");
        long after=new java.util.Date().getTime();
        System.out.println("Parsing: "+(after-before));

        QueryExecution qexec =
QueryExecutionFactory.sparqlService(OntologyAcces.HOST_QUERY, query);

        ResultSet results = null;

        try{
            long beforeE=new java.util.Date().getTime();
            results = qexec.execSelect();
            long afterE=new java.util.Date().getTime();
            System.out.println("Execution: "+(afterE-beforeE));

//          ResultSetFormatter.out(System.out, results, query) ;
            for(;results.hasNext();){
                QuerySolution qs = results.nextSolution();
                //System.out.println("I'm the subject: "+qs.getResource("s"));

            }
            qexec.close();
        } catch (Exception e) {
            e.printStackTrace(); }


The parsing is taking up to 1000 ms and te execution is taking up to 200
ms. The triple store answer is not taking much time, the trouble I have is
with ARQ specifically.
-- 
*FEDERICO LÓPEZ GÓMEZ*
Estudiante Ingeniería de Sistemas
Universidad del Valle - Sede Tuluá

Re: ARQ Takes to much time to parse a query.

Posted by Federico López <fi...@gmail.com>.
I run it several times like you did it in a for sentence and the second
time was much better like you said.
Thanks a lot. Now I understand what yuo said about class loading.

2012/6/10 Federico López <fi...@gmail.com>

> I'm using arq.2.8.8 in a Intel Core 2 Quad (2.33 Ghz) with   4.00 GB RAM
> in Windows 7 x 64.
>
> I'm using Eclipse Indigo IDE.
>
>
> OntologyAcces.HOSTQUERY="http://127.0.0.1:3030/tdb/query" is a constant
> that keeps the value of the URL of the Fuseki server.
>
>
> 2012/6/10 Andy Seaborne <an...@apache.org>
>
>> On 10/06/12 18:36, Federico López wrote:
>>
>>> I need to send a query to Fuseki server from Java ans ARQ is taking too
>>> much time parsing and executing a simple query to Fuseki Server. This is
>>> my
>>> code:
>>>
>>> long before=new java.util.Date().getTime();
>>>         Query query =
>>> QueryFactory.create(**OntologyAcces.PREFIXES+"SELECT *{?s ?p ?o} LIMIT
>>> 1000");
>>>         long after=new java.util.Date().getTime();
>>>         System.out.println("Parsing: "+(after-before));
>>>
>>
>> If this is the first time the query is parsed, java classloading will
>> happen.
>>
>> I put that code (minus OntologyAcces.PREFIXES) in a loop executes 10 time
>> and got:
>>
>> Parsing: 28
>> Parsing: 1
>> Parsing: 0
>> Parsing: 1
>> Parsing: 0
>> Parsing: 1
>> Parsing: 0
>> Parsing: 1
>> Parsing: 0
>> Parsing: 0
>>
>> as you can see, the first time is expensive.
>>
>> Are you using a laptop?
>>
>>
>>
>>
>>>         QueryExecution qexec =
>>> QueryExecutionFactory.**sparqlService(OntologyAcces.**HOST_QUERY,
>>> query);
>>>
>>
>> What's OntologyAcces.HOST_QUERY
>>
>>
>>
>>>         ResultSet results = null;
>>>
>>>         try{
>>>             long beforeE=new java.util.Date().getTime();
>>>             results = qexec.execSelect();
>>>             long afterE=new java.util.Date().getTime();
>>>             System.out.println("Execution: "+(afterE-beforeE));
>>>
>>> //          ResultSetFormatter.out(System.**out, results, query) ;
>>>             for(;results.hasNext();){
>>>                 QuerySolution qs = results.nextSolution();
>>>                 //System.out.println("I'm the subject:
>>> "+qs.getResource("s"));
>>>
>>>             }
>>>             qexec.close();
>>>         } catch (Exception e) {
>>>             e.printStackTrace(); }
>>>
>>>
>>> The parsing is taking up to 1000 ms and te execution is taking up to 200
>>> ms. The triple store answer is not taking much time, the trouble I have
>>> is
>>> with ARQ specifically.
>>>
>>
>> 200ms to execute to a HTTP operation to a server without a pre-existing
>> connection is quite reasonable.  There are lots of things that might be
>> being done.
>>
>>        Andy
>>
>>
>
>
> --
> *FEDERICO LÓPEZ GÓMEZ*
> Estudiante Ingeniería de Sistemas
> Universidad del Valle - Sede Tuluá
>
>


-- 
*FEDERICO LÓPEZ GÓMEZ*
Estudiante Ingeniería de Sistemas
Universidad del Valle - Sede Tuluá

Re: ARQ Takes to much time to parse a query.

Posted by Federico López <fi...@gmail.com>.
I'm using arq.2.8.8 in a Intel Core 2 Quad (2.33 Ghz) with   4.00 GB RAM in
Windows 7 x 64.

I'm using Eclipse Indigo IDE.


OntologyAcces.HOSTQUERY="http://127.0.0.1:3030/tdb/query" is a constant
that keeps the value of the URL of the Fuseki server.


2012/6/10 Andy Seaborne <an...@apache.org>

> On 10/06/12 18:36, Federico López wrote:
>
>> I need to send a query to Fuseki server from Java ans ARQ is taking too
>> much time parsing and executing a simple query to Fuseki Server. This is
>> my
>> code:
>>
>> long before=new java.util.Date().getTime();
>>         Query query =
>> QueryFactory.create(**OntologyAcces.PREFIXES+"SELECT *{?s ?p ?o} LIMIT
>> 1000");
>>         long after=new java.util.Date().getTime();
>>         System.out.println("Parsing: "+(after-before));
>>
>
> If this is the first time the query is parsed, java classloading will
> happen.
>
> I put that code (minus OntologyAcces.PREFIXES) in a loop executes 10 time
> and got:
>
> Parsing: 28
> Parsing: 1
> Parsing: 0
> Parsing: 1
> Parsing: 0
> Parsing: 1
> Parsing: 0
> Parsing: 1
> Parsing: 0
> Parsing: 0
>
> as you can see, the first time is expensive.
>
> Are you using a laptop?
>
>
>
>
>>         QueryExecution qexec =
>> QueryExecutionFactory.**sparqlService(OntologyAcces.**HOST_QUERY, query);
>>
>
> What's OntologyAcces.HOST_QUERY
>
>
>
>>         ResultSet results = null;
>>
>>         try{
>>             long beforeE=new java.util.Date().getTime();
>>             results = qexec.execSelect();
>>             long afterE=new java.util.Date().getTime();
>>             System.out.println("Execution: "+(afterE-beforeE));
>>
>> //          ResultSetFormatter.out(System.**out, results, query) ;
>>             for(;results.hasNext();){
>>                 QuerySolution qs = results.nextSolution();
>>                 //System.out.println("I'm the subject:
>> "+qs.getResource("s"));
>>
>>             }
>>             qexec.close();
>>         } catch (Exception e) {
>>             e.printStackTrace(); }
>>
>>
>> The parsing is taking up to 1000 ms and te execution is taking up to 200
>> ms. The triple store answer is not taking much time, the trouble I have is
>> with ARQ specifically.
>>
>
> 200ms to execute to a HTTP operation to a server without a pre-existing
> connection is quite reasonable.  There are lots of things that might be
> being done.
>
>        Andy
>
>


-- 
*FEDERICO LÓPEZ GÓMEZ*
Estudiante Ingeniería de Sistemas
Universidad del Valle - Sede Tuluá

Re: ARQ Takes to much time to parse a query.

Posted by Andy Seaborne <an...@apache.org>.
On 10/06/12 18:36, Federico López wrote:
> I need to send a query to Fuseki server from Java ans ARQ is taking too
> much time parsing and executing a simple query to Fuseki Server. This is my
> code:
>
> long before=new java.util.Date().getTime();
>          Query query =
> QueryFactory.create(OntologyAcces.PREFIXES+"SELECT *{?s ?p ?o} LIMIT
> 1000");
>          long after=new java.util.Date().getTime();
>          System.out.println("Parsing: "+(after-before));

If this is the first time the query is parsed, java classloading will 
happen.

I put that code (minus OntologyAcces.PREFIXES) in a loop executes 10 
time and got:

Parsing: 28
Parsing: 1
Parsing: 0
Parsing: 1
Parsing: 0
Parsing: 1
Parsing: 0
Parsing: 1
Parsing: 0
Parsing: 0

as you can see, the first time is expensive.

Are you using a laptop?


>
>          QueryExecution qexec =
> QueryExecutionFactory.sparqlService(OntologyAcces.HOST_QUERY, query);

What's OntologyAcces.HOST_QUERY

>
>          ResultSet results = null;
>
>          try{
>              long beforeE=new java.util.Date().getTime();
>              results = qexec.execSelect();
>              long afterE=new java.util.Date().getTime();
>              System.out.println("Execution: "+(afterE-beforeE));
>
> //          ResultSetFormatter.out(System.out, results, query) ;
>              for(;results.hasNext();){
>                  QuerySolution qs = results.nextSolution();
>                  //System.out.println("I'm the subject: "+qs.getResource("s"));
>
>              }
>              qexec.close();
>          } catch (Exception e) {
>              e.printStackTrace(); }
>
>
> The parsing is taking up to 1000 ms and te execution is taking up to 200
> ms. The triple store answer is not taking much time, the trouble I have is
> with ARQ specifically.

200ms to execute to a HTTP operation to a server without a pre-existing 
connection is quite reasonable.  There are lots of things that might be 
being done.

	Andy