Hello everybody,
First, excuse me for the mistakes I may have made.
I am a trainee and the enterprise I work for is planning to
change the presentation layer (and maybe all the layers) of their
software packages for Flex (at the present time they use the .NET
platform). But in order to ensure that Flex can answer their needs,
they have asked me to make a model using Flex together with Java,
Hibernate, LCDS, SQL Server and Tomcat.
I am currently using those versions :
- jdk 1.5.0_12
- Hibernate 3 (I use the jars provided with LCDS)
- LCDS 2.5
- SQL Server 2005
- Tomcat 5.5
As far as some of the tables from their databases are quite
big (they can reach 2.000.000 rows) I need to implement paging. The
table I currently use for my example is "only" 10.000 rows.
I must confess that I am new to n-tiers architectures and
that there are many concepts I misunderstand.
What I want to do is make the user able to choose how many
rows a page he wants to display (This could look a bit like
this). And
the server would only send this number of rows for each page (which
is not the case for the page I linked).
I have noticed that the implementation of the fill method for
the HibernateAssembler had changed between FDS (
public Collection fill(List fillArgs)) and LCDS (
public Collection fill(List fillArgs, int startIndex, int
numItems)).
You are now able to specify a start index and the number of
rows to display. This is the method I would like to use.
To do so, I have tried to refer to
LCDS
documentation but I find it quite confusing. Here it is :
Using on-demand pagingThere are two ways in which the Data Management Service
implements on-demand paging of data by Flex clients. In the default
approach, the fill method returns the entire collection of objects.
The server sends the first page of items to the client as part of
the initial fill request.
As the client code tries to access ArrayCollection elements
which are not resident, additional pages are retrieved from the
server. If the cache-items property is set to true, these pages are
sent directly by the Data Management Service. If the cache-items
property is set to false, the Data Management Service calls the
getItem() method for each item it needs to send to the client.
When the original query is too large to be held in the
server's memory, it is desirable for the Assembler to only return a
single page of items at a time. In this mode, each time the client
requests a page, the Assembler is asked for that page of items and
they are sent directly to the client. Currently this mode is not
supported when the autoSyncEnabled property is set to true for the
paged collection. You must set the autoSyncEnabled property to
false before executing the code. This configuration is not
supported for the SQLAssembler. It is supported for the
HibernateAssembler and you can implement it with your own Java
Assembler implementation.
To enable this second mode, you set the custom="true"
attribute on the paging element in your network settings on the
server and ensure paging is enabled with a reasonable pageSize
value.
If you are using the fill-method approach in your Assembler,
you specify a fill method that matches the fill parameters used by
your client with two additional parameters: the startIndex and the
number of items requested. The Data Management Service calls this
method once for each page requested by the client.
If you are using the Assembler interface approach, you
override the useFillPage() method for the appropriate fill method
to return true for the fill parameters you want paged using custom
paging. When the client requests a page, it calls the fill variant
in the Assembler interface which takes the additional startIndex
and number of items parameters. You return
just those items and they are returned to the client.
When you use custom paging in either the fill-method or
Assembler approach, there are two ways that the client can
determine the size of the collection. When you have enabled the
custom paging mode, on the initial fill request made by the client,
the Data Management Service invokes the count method with the
parameters the client passed to fill. If the count
method returns -1, a dynamic sizing algorithm is used for
this filled collection. In this approach, the assembler's paged
fill method is called with startIndex of 0 and number of items set
to the pageSize + 1. If the assembler method returns less than the
number requested, the size of the fill is known. If it returns the
pageSize+1 items requested, pageSize items are
returned to the client but the client sets the collection
size to pageSize + 1 - with one empty slot at the end. When that
empty item is requested by the client, the next pageSize+1 items
are requested and the process repeats until the assembler returns
less than pageSize+1 items.
The HibernateAssembler implements this dynamic paging
mechanism when you set the page-queries-from-database attribute to
true in the destination's server properties. If you are using an
HQL query sent from the client and the query is a simple query, the
Hibernate assembler attempts to implement a count query by
modifying the query sent from the client.
If you are using a named query, the Hibernate assembler looks
for a query named <original-query-name>.count. If that query
exists, it uses it to compute the size of the paged collection.
Otherwise, it uses the dynamic-sized approach.
I guess I should change
<params>java.util.List</params> for
<params>java.util.List,int,int</params>.
But I don't know which properties to add or modify.
It also talks about the custom attribute of the paging
element. Is it a new attribute that replaces "enabled" ? Or does it
come in addition to it ?
The following files are working and allow me to retrieve the
datas I expect (but without paging).
My data-management-config.xml looks like this :
<?xml version="1.0" encoding="UTF-8"?>
<service id="data-service" class="flex.data.DataService"
messageTypes="flex.data.messages.DataMessage">
<adapters>
....<adapter-definition id="java-adapter"
class="flex.data.adapters.JavaAdapter" default="true"/>
</adapters>
<default-channels>
....<channel ref="my-rtmp"/>
</default-channels>
<destination id="refacteur.hibernate">
....<adapter ref="java-adapter" />
....<properties>
........<use-transactions>true</use-transactions>
........<source>flex.data.assemblers.HibernateAssembler</source>
........<scope>application</scope>
........<metadata>
............<identity property="cleacteur"/>
........</metadata>
........<network>
............<session-timeout>20</session-timeout>
............<paging enabled="false" pageSize="20"/>
............<throttle-inbound policy="ERROR"
max-frequency="500"/>
............<throttle-outbound policy="REPLACE"
max-frequency="500"/>
........</network>
........<server>
............<hibernate-entity>fr.phylum.referentiel.Refacteur</hibernate-entity>
............<fill-method>
................<name>fill</name>
................<params>java.util.List</params>
............</fill-method>
............<fill-configuration>
................<use-query-cache>true</use-query-cache>
................<allow-hql-queries>true</allow-hql-queries>
............</fill-configuration>
........</server>
....</properties>
</destination>
</service>
This is the named query definition I use, found in
Refacteur.hbm.xml :
<query name="refacteur.where.ville.order.by.libacteur">
....<![CDATA[
........from Refacteur as refacteur
........where refacteur.villeActeur = ?
........order by libacteur
....]]>
</query>
And here is the call to the fill method, found in my mxml
application file Acteurs.mxml :
globalFilter.dataService.fill(globalFilter.arrayCollection,
"refacteur.where.ville.order.by.libacteur", ["VANNES"]);
I have tried to modify several things, but nothing works.
I would be most grateful if someone could tell me what to
change in my files in order to make the paging work.
Nathalie.
