Many thanks for adding this feature. Without it the Clojure code would
have been left in the dust.
On Aug 9, 11:46 pm, Sean Corfield wrote:
> On Tue, Aug 9, 2011 at 9:39 AM, Shoeb Bhinderwala <
>
> shoeb.bhinderw...@gmail.com> wrote:
> > With these options added the Clojure code runs just about a
On Tue, Aug 9, 2011 at 9:39 AM, Shoeb Bhinderwala <
shoeb.bhinderw...@gmail.com> wrote:
> With these options added the Clojure code runs just about as fast as
> Java. I set the fetch size to 1000 for both of them.
>
> Average run times to load 69,000 records:
>
> Java = 2.67 seconds
> Clojure =
Nice work getting to the bottom of this.
Sent from my iPad
On 9 Aug 2011, at 17:39, Shoeb Bhinderwala wrote:
> Hi Sean –
>
> With these options added the Clojure code runs just about as fast as
> Java. I set the fetch size to 1000 for both of them.
>
> Average run times to load 69,000 records:
Hi Sean –
With these options added the Clojure code runs just about as fast as
Java. I set the fetch size to 1000 for both of them.
Average run times to load 69,000 records:
Java = 2.67 seconds
Clojure = 2.72 seconds
Thanks
Shoeb
On Aug 9, 12:54 am, Sean Corfield wrote:
> On Mon, Aug 8,
> So {:result-type :forward-only :fetch-size 1000} in the vector before your
> SQL string should provide the same optimization as in the Java code. Please
> let us know the results of your next test with those options added.
As per this post http://blog.lishman.com/2008/03/jdbc-fetch-size.html
act
On Mon, Aug 8, 2011 at 5:12 PM, Dmitry Gutov wrote:
> Like the doc says, `with-query-results` takes additional options
> before the query string.
>
Yup, this is a recent enhancement. You can either pass a PreparedStatement
that you've built and configured instead of a SQL string or you can pass
Like the doc says, `with-query-results` takes additional options
before the query string.
In your case, it should be something like this:
(sql/with-query-results recs
`[{:fetch-size 1000} ~query-str ~@bvs]
(doall recs))
--
You received this message because you are subscribed to the Google
Gr
Sean/Stuart/Others -
My apologies to the group. I found out why my Clojure code runs slower
than Java.
The Java code uses the setFetchSize() method to retrieve data in
batch:
myResultSet.setFetchSize(1000);
myResultSet.setFetchDirection(ResultSet.FETCH_FORWARD);
Without
You are right Michael. I misunderstood Colin's statement.
As Stuart suggested I am profiling the code and will share the results
with the group soon.
On Aug 8, 4:33 am, Michael Wood wrote:
> Hi Shoeb
>
> On 7 August 2011 01:51, Shoeb Bhinderwala wrote:
>
> > I am not guessing. I measured the pe
Hi Shoeb
On 7 August 2011 01:51, Shoeb Bhinderwala wrote:
> I am not guessing. I measured the performance of the query using plain
> Java and Clojure.
That's not what he was saying. In your first message you said:
>>> I am assuming that most of the extra time is spent converting the results
>>
Thank you for your information! I think I will start with c.j.j's test
code :)
On Aug 7, 3:08 pm, Sean Corfield wrote:
> On Sat, Aug 6, 2011 at 11:04 PM, jaime wrote:
> > Hi guys, I just want to learn using databases in Clojure, can you
> > suggest where to start? by looking at source code o
Hi Shoeb,
At this point, we probably need more data to give a meaningful answer.
Different core data structures and different coding conventions can mean
that "the same code" in Clojure and Java will behave quite differently.
Try profiling your Clojure code with a Java profiler such as VisualVM
On Sat, Aug 6, 2011 at 11:04 PM, jaime wrote:
> Hi guys, I just want to learn using databases in Clojure, can you
> suggest where to start? by looking at source code of clojure.java.jdbc/
> clojure.contrib.sql or there's some tutorial/document that I can start
> with?
>
The tests in c.j.j are pr
On Sat, Aug 6, 2011 at 10:51 PM, Shoeb Bhinderwala <
shoeb.bhinderw...@gmail.com> wrote:
> I switched to clojure.java.jdbc. Found no difference at all. It is
> still about 10 times slower than java.
>
Thanx. I wanted to eliminate reflection as the potential culprit.
Have you tried Clojure 1.3, j
On Aug 7, 10:51 am, Shoeb Bhinderwala
wrote:
> I switched to clojure.java.jdbc. Found no difference at all. It is
> still about 10 times slower than java.
I guess you can now get c.j.jdbc and put (time ...) wrappers in the
call path in it to see where is the time mostly spent.
Do share your fi
Hi guys, I just want to learn using databases in Clojure, can you
suggest where to start? by looking at source code of clojure.java.jdbc/
clojure.contrib.sql or there's some tutorial/document that I can start
with?
Thanks,
Jaime
On Aug 7, 1:51 pm, Shoeb Bhinderwala
wrote:
> I switched to clojure
I switched to clojure.java.jdbc. Found no difference at all. It is
still about 10 times slower than java.
On Aug 6, 8:54 pm, Sean Corfield wrote:
> On Sat, Aug 6, 2011 at 4:51 PM, Shoeb Bhinderwala <
>
> shoeb.bhinderw...@gmail.com> wrote:
> > In one test case, I loaded 69,099 records. The Java c
On Sat, Aug 6, 2011 at 4:51 PM, Shoeb Bhinderwala <
shoeb.bhinderw...@gmail.com> wrote:
> In one test case, I loaded 69,099 records. The Java code took 5
> seconds to execute the query and create as many objects. Clojure code
> took 50.43 seconds.
>
Try using clojure.java.jdbc instead of clojure.
I am not guessing. I measured the performance of the query using plain
Java and Clojure.
In one test case, I loaded 69,099 records. The Java code took 5
seconds to execute the query and create as many objects. Clojure code
took 50.43 seconds.
The JVM settings are the same - both are initialized t
That assumption needs checking - first rule of performance analysis: check,
don't guess :)
For example, is the java code using an existing connection versus clojure
creating one? I would also time the cost of creating 10 clojure maps of
a similar structure. Finally - 100,000 is big enough to
20 matches
Mail list logo
