W3C home > Mailing lists > Public > semantic-web@w3.org > February 2006

Re: Is it just me or does this seem incredibly slow?

From: Garrett Wollman <wollman+semantic-web@bimajority.org>
Date: Sun, 5 Feb 2006 20:23:47 -0500
Message-ID: <17382.42275.629537.946500@khavrinen.csail.mit.edu>
To: Dave Beckett <dave@dajobe.org>
Cc: semantic-web@w3.org

<<On Sun, 05 Feb 2006 15:14:52 -0800, Dave Beckett <dave@dajobe.org> said:

> The web demo uses rasqal 0.9.11 with redland on a memory based store
> (with no indexing) so it's unlikely to be that.  5 minute queries
> usually means something went wrong, and as you don't give the full
> query, I'm not clear what it could be.

No, actually, what I wrote *was* the whole query -- I've been testing
on simple queries on the theory that they ought to be faster than the
more complex queries that I haven't translated into SPARQL yet.  The
five-minute time was for the same query executed with rdfproc query
versus one minute for roqet.  (The one-minute time with cwm --strings
was to extract all the information I need, on the theory that I would
have to stick it into my RDBMS to get acceptable performance.
However, cwm was only working with a fragment of the store.)

> Also DISTINCT had bug fixes and improvements in rasqal 0.9.11 so I
> assume you are using that.

No, I'm using 0.9.10 which is what comes with the current FreeBSD
package.  I was going to file a bug report that DISTINCT was broken;
I'm glad to see it's already been fixed!

With a hint from Richard Newman I was able to get the query to run 46%
faster rewriting the query slightly, but that's still far too slow:

--- foo.sparql.orig     Sun Feb  5 19:59:53 2006
+++ foo.sparql  Sun Feb  5 19:51:57 2006
@@ -15,9 +15,9 @@
 SELECT DISTINCT ?photo, ?name
 WHERE
 {
-       ?photo rdf:type photo:ImageFile ;
-              foaf:depicts ?b .
-       ?b rdf:type foaf:Person ;
-          foaf:name ?name
+       ?photo rdf:type photo:ImageFile .
+       ?photo foaf:depicts ?b .
+       ?b rdf:type foaf:Person .
+       ?b foaf:name ?name
 }
 ORDER BY ?name

But the spec says these forms are absolutely identical!  (I also tried
removing the comma from the variable list, which is apparently no
longer in the spec even though it is in all the examples I've seen; it
makes no difference in the speed of the query.)

I'm now off to see if I can upgrade the port to the latest version and
whether that makes any difference.

-GAWollman
Received on Monday, 6 February 2006 01:23:54 UTC

This archive was generated by hypermail 2.4.0 : Thursday, 24 March 2022 20:40:58 UTC