Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [eclipselink-users] MySql problem with ReadAllQuery and pagination

Hello Sonavor,

If you use a readAllQuery and do not specify the firstResult/maxRows, then all the results are returned. This means that the query is executed, all results are sent over the connection, and all the objects are built and cached, so there is no reason to then use pagination on multiple queries - you can just use the result list already returned and iterate over it instead. I would recommend using a ReportQuery instead of ReadAllQuery to count the number of rows. This would allow you to reuse the expression, and just get a different result instead;

ReportQuery reportQuery = new ReportQuery(MediaListing.class);
reportQuery.setSelectionCriteria(exp);
reportQuery.count(reportQuery.getExpressionBuilder());
result = em.getDelegate().getActiveSession().executeQuery(reportQuery);

This is essentially the same as executing a JPQL count query. I haven't looked into executing it through the JPA interface, but you may need to use setShouldReturnWithoutReportQueryResult(true) to get the correct type.
Best Regards,
Chris



sonavor wrote:
Cool, Chris.  Thanks.

I like your solution.  I wasn't aware of the code -

javax.persistence.Query jpaquery =
((JpaEntityManager)em.getDelegate()).createQuery(readAllQuery);

I tried it out and it works great.

One other question I have regarding pagination involving the ReadAllQuery. I usually try to just use the javax.persistence.Query with JPA QL. I only
use ExpressionBuilder with ReadAllQuery when I have a query that is joining
a many-to-many relationship along with additional relationship tables
(usually an advanced search screen type situation). That is the case here. My question is about getting the total result count for the query
pagination.  The way I am doing it now is I run the same ReadAllQuery
without any max row setting and capture the size of the result set.  I only
do that once per query submit.  After that the pagination mechanism just
calls the query with the starting row and max row value.

Is that the only way of solving the get count when a ReadAllQuery is
involved?  In cases where I use a javax.persistence.Query I just create a
"count" query ( SELECT count(o) FROM MyJpaClass as o WHERE .... whatever
other query attributes are involved ).

Thanks again,
sonavor



christopher.delahunt wrote:
Hello,

While this is counter intuitive when using pagination, EclipseLink's maxRows and firstResult is based on using the statement to set the maxRows to return and then the resultset after the query executes to jump to the firstResult indicated. The MaxRows is applied before the firstResult. This means that values of 4 for maxrows and then 5 for the first result that it would skip to the 5th record when there are only 4. The maxRow value needs to be 9, with firstResult set to 5 to get 4 rows back.

When pagination was implemented, it used the same maxRows but had to correct for this when the value is passed to the limit function in MySQL, which is why you see a -1. To use the native maxRows with firstResult, you will need to add the first rows value to the maxResults.

You will see different behaviour when using the JPA Query setMaxResults than when using the native setMaxRows. You can use

ReadAllQuery readAllQuery = new ReadAllQuery(MediaListing.class);
readAllQuery.setSelectionCriteria(exp);
readAllQuery.prepareForExecution();
javax.persistence.Query jpaquery =
((JpaEntityManager)em.getDelegate()).createQuery(readAllQuery);
query.setMaxResults(rowCount);
query.setFirstResult(startRow);
List<MediaListing> results = query.getResultList();


Or you can use the createQuery(Expression, Class) method and avoid creating the ReadAllQuery.
Best Regards,
Chris

sonavor wrote:
I have a MySql database (v5.1) where I created a small database
consisting of
several tables that use a variety of relationships (one-one, one-many,
many-to-many).  In testing how to provide a user with multiple search
options I decided to use the EclipseLink ExpressionBuilder.  This was
successful and I tested a bunch of search criteria combinations.

My ExpressionBuilder solution is used with a ReadAllQuery as follows:

ReadAllQuery readAllQuery = new ReadAllQuery(MediaListing.class);
readAllQuery.setSelectionCriteria(exp);
readAllQuery.setFirstResult(0);
readAllQuery.prepareForExecution();
Session session = ((org.eclipse.persistence.jpa.JpaEntityManager)
em.getDelegate()).getActiveSession();
List<MediaListing> results = (List) session.executeQuery(readAllQuery);

Once I was satisfied with the search results I got with this query I
moved
on to implement pagination of the search results.  That is where I have
run
into problems.

My updated ReadQllQuery for pagination is this -

ReadAllQuery readAllQuery = new ReadAllQuery(MediaListing.class);
readAllQuery.setSelectionCriteria(exp);
readAllQuery.setFirstResult(startRow);
readAllQuery.setMaxRows(rowCount);
readAllQuery.prepareForExecution();
Session session = ((org.eclipse.persistence.jpa.JpaEntityManager)
em.getDelegate()).getActiveSession();
List<MediaListing> results = (List) session.executeQuery(readAllQuery);

When I run it the first time, startRow = 0 and rowCount = 4
The EclipseLink logging shows this (the first page of the result set) -

Fine]: 2009-08-04
23:34:30.018--ServerSession(30709456)--Connection(19936982)--Thread(Thread[TP-Processor3,5,main])--SELECT
MEDIA_ID AS MEDIA_ID1, LAST_UPDATE AS LAST_UPDATE2, MEDIA_CODE AS
MEDIA_CODE3, MEDIA_RATING AS MEDIA_RATING4, TOTAL_PURCH_COUNT AS
TOTAL_PURCH_COUNT5, MEDIA_TITLE AS MEDIA_TITLE6, MEDIA_DESCRIPTION AS
MEDIA_DESCRIPTION7, PREDOMINATE_COLOR AS PREDOMINATE_COLOR8, VIEW_COUNT
AS
VIEW_COUNT9, AVAILABILITY AS AVAILABILITY10, MEDIA_CATEGORY_ID AS
MEDIA_CATEGORY_ID11, AUTHOR_ID AS AUTHOR_ID12, MEDIA_TYPE_ID AS
MEDIA_TYPE_ID13, MEDIA_SUBJECT_ID AS MEDIA_SUBJECT_ID14, MEDIA_FORMAT_ID
AS
MEDIA_FORMAT_ID15 FROM MEDIA_LISTING LIMIT ?, ?
        bind => [0, 4]

Notice the bind has the starting value of 0 and the maximum number of
rows
to return as 4

When I click on my pagination control to go to page 2 the submitted
startRow
value = 5 and the rowCount = 4
However, EclipseLink bombs with -

Exception [EclipseLink-4002] (Eclipse Persistence Services -
1.1.2.v20090612-r4475):
org.eclipse.persistence.exceptions.DatabaseException
Internal Exception:
com.mysql.jdbc.exceptions.jdbc4.MySQLSyntaxErrorException: You have an
error
in your SQL syntax; check the manual that corresponds to your MySQL
server
version for the right syntax to use near '-1' at line 1
Error Code: 1064
Call: SELECT MEDIA_ID AS MEDIA_ID1, LAST_UPDATE AS LAST_UPDATE2,
MEDIA_CODE
AS MEDIA_CODE3, MEDIA_RATING AS MEDIA_RATING4, TOTAL_PURCH_COUNT AS
TOTAL_PURCH_COUNT5, MEDIA_TITLE AS MEDIA_TITLE6, MEDIA_DESCRIPTION AS
MEDIA_DESCRIPTION7, PREDOMINATE_COLOR AS PREDOMINATE_COLOR8, VIEW_COUNT
AS
VIEW_COUNT9, AVAILABILITY AS AVAILABILITY10, MEDIA_CATEGORY_ID AS
MEDIA_CATEGORY_ID11, AUTHOR_ID AS AUTHOR_ID12, MEDIA_TYPE_ID AS
MEDIA_TYPE_ID13, MEDIA_SUBJECT_ID AS MEDIA_SUBJECT_ID14, MEDIA_FORMAT_ID
AS
MEDIA_FORMAT_ID15 FROM MEDIA_LISTING LIMIT ?, ?
        bind => [5, -1]

Notice the bind shows the correct start row of 5 but for some reason the
maximum row count is -1 !!!!

Does anyone know what could be causing that?  I put in a debug statement
and
verified that I was setting the readAllQuery.setMaxRows(rowCount); with a
rowCount value of 4.
As a sanity check I created a similar scenario with a test table called
TEST_LISTING and tried my pagination control with a JPA Query provided by
the EntityManager.  It is defined like this -

Query q = em.createNamedQuery("TestListing.findByGroupingCode");
q.setParameter("groupingCode", groupCode);
q.setFirstResult(startingRow);
q.setMaxResults(rowCount);
List<TestListing> tList = q.getResultList();

In the test with the Query object my pagination works without any
problems.

Is there a bug with the ReadAllQuery and setMaxRows ?  Or is it something
to
do just with MySql and the ReadAllQuery?  Maybe some problem with the
MySql
JDBC driver?

As one last test I ran the EclipseLink generated SQL -
SELECT MEDIA_ID AS MEDIA_ID1, LAST_UPDATE AS LAST_UPDATE2, MEDIA_CODE AS
MEDIA_CODE3, MEDIA_RATING AS MEDIA_RATING4, TOTAL_PURCH_COUNT AS
TOTAL_PURCH_COUNT5, MEDIA_TITLE AS MEDIA_TITLE6, MEDIA_DESCRIPTION AS
MEDIA_DESCRIPTION7, PREDOMINATE_COLOR AS PREDOMINATE_COLOR8, VIEW_COUNT
AS
VIEW_COUNT9, AVAILABILITY AS AVAILABILITY10, MEDIA_CATEGORY_ID AS
MEDIA_CATEGORY_ID11, AUTHOR_ID AS AUTHOR_ID12, MEDIA_TYPE_ID AS
MEDIA_TYPE_ID13, MEDIA_SUBJECT_ID AS MEDIA_SUBJECT_ID14, MEDIA_FORMAT_ID
AS
MEDIA_FORMAT_ID15 FROM MEDIA_LISTING LIMIT ?, ?

In a MySql query window and tried the various startRow and rowCount
values
manually.  The MySql query window executed the query and returned the
expected results.

Thanks for any help with this matter.




_______________________________________________
eclipselink-users mailing list
eclipselink-users@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/eclipselink-users





Back to the top