ADF Best Practices

Collated by: Sarah Maria Justina & Akhil Rangaraju

Oracle OMS

Target Audience
This document will be a helpful reference point for ADF developers,
ADF Tech Leads and ADF Solution architects. This document
assumes a good understanding of ADF and a conceptual know-how
of J2EE framework.

Overview

 This document encompasses some of the best practices that must be

followed when designing and building an ADF application. The
document talks about these best practices in the following sequence
1. Entity Object
2. View Object
3. Application Module
4. Other General ADF BC Best Practices
5. Pages
6. JavaScript
7. Bindings
8. Taskflows
9. Beans
10. Other General Java Best Practices
11. web.xml
12. weblogic.xml
13. Deployment

Entity Objects

Define at least one entity object key attribute per entity object
- An entity object must has at least one attribute defined as the
key.
Do not access an application module from an entity object entity objects should not access application modules as this
makes the entity object only useable within that application
module.
Don't use ROWID as entity object key - By default if ADF BC
doesn't find a primary key at design time for an entity object's
related database table, it will generate an additional attribute
mapping to ROWID. This will continue to work unless an Oracle
database Real Application Cluster(RAC) is used where across
database partitions different ROWIDs can be returned for the
same row.
Define entity object attribute labels for UI displayed
attributes - For all entity object attributes that are displayed in the
UI set their default label.

Ensure entity object UI Hints and messages are

internationalized Assuming that the application will be
translated all entity object field labels and other relevant UI hints
and messages need to be managed via message bundles.
Use entity object attribute change indicators where
appropriate - When creating an entity object that includes a
column that indicates when the row was created and updated, set
the change indicator property for that attribute. This provides a
performance improvement when updating or deleting the record to
the database as ADF BC only needs to check the attribute as
marked with the change indicator rather than all attributes of the
entity.
Define entity object attribute tooltips for UI displayed
attributes where an explanation of the allowable values is
required - For all entity object attributes that are displayed in the
UI, where the attribute has a set of allowable value that isn't
obvious, include a one sentence tooltip to describe this.

Use entity object history columns where appropriate - When

creating an entity object based on a table with columns such as
created_by, created_date, updated_by and updated_date, set the
relating attribute history column properties.
Create error messages for entity object mandatory and
precision validators The default error messages for the pregenerated mandatory and precision validators can be overridden
with clearer messages.
Avoid using the entity object validateEntity() method for
validation - do not override an entity object's EntityImpl
validateEntity() method. Rather code any business logic as a
declarative method validator for the entity object itself. This comes
with the added advantage all validations are visible in the entity
object's business rules page.

Avoid entity object attribute validation in accessors - do not

write code to validate an attribute in its setter accessor within the
associated EntityImpl class for the entity object. Rather create a
declarative method validator for the attribute.
Do not call postChanges unless you can guarantee a commit
or rollback at the end the HTTP request Typically the
EntityImpl postChanges method is called by the ADF BC lifecycle
upon a commit operation. However you can also programmatically
call this method. As the postChanges method changes the state of
the associated EntityImpl which then assumes the changes have
been committed to the database, this can leave your entity object
in a dirty state. As such only call the postChanges method directly
if you know that at the end of the HTTP request you will commit or
rollback the transaction.

View Objects



Never use view object select * - Never use SELECT * for a view
object query. Always specify all columns in the query.
Avoid hardcoding view object queries with literals - Don't
hardcode view object where clauses or view criteria with literals.
Always use bind variables.
Use view object bind variables over dynamic SQL - Bind
variables also make the queries of view objects less susceptible to
SQL injection attacks.
Don't define circular list of value dependencies - View objects
allow you to define dependent/chained list of values. Be careful
not to define a circular dependency as this can result in an infinite
loop.
Ensure view object UI Hints and messages internationalized Assuming that the application will be translated all view object
field labels and other relevant UI hints and messages need to be
managed via message bundles.

Avoid unnecessary view object duplication - In many cases

very similar view object instances can be created from the same
base definition with the correct use of applied view criteria when
the instance is exposed via the application module.
View objects should include all key attributes from underlying
entity objects If a view object is based on one or more entity
objects, then view object should include all the key attributes from
each entity object.
Define at least one view object key attribute - A view object
must has at least one attribute defined as the key. This is
particularly important for view objects that are marked for
passivation as the key is required to restore the current row on
activation.
Rarely use view object expert mode - Rarely use expert mode
for view objects. If the query you wish to execute can be
implemented by modifying other properties of the view object over
expert mode this should be your preference. For example SQL
functions on attributes can be applied at the attribute level.

Use declarative SQL view objects where possible - Declarative

SQL based view objects selectively prune the attributes included
in the underlying query based on the attributes accessed via the
binding and UI layers. This results in faster turn around with the
database and less middle tier memory consumed.
Avoid read only view objects for speed benefits alone - Older
ADF manuals recommended using read only view objects as
there was a minor speed benefit. Since JDeveloper 11g this is no
longer true as the caching mechanism has been improved. Today
read only view objects should be reserved for complex queries
such as those with set operations (e.g. UNION, MINUS,
INTERSECT) and connect by queries. All other view objects
should be based on entity objects. If the view object needs to be
read only, unselect the "Updateable" property when the entity
object is mapped to view objects.

Avoid case insensitive view criteria if not needed - For view

objects using view criteria, ensure to uncheck the case insensitive
option if it not needed as it will make database queries slower
unless an Oracle RDBMS function based index is available.
Ensure to use view object accessor tuning - ensure to set the
view object exposed through the view accessor tuning options
appropriate to how the view object will be used.
Use type-safe view object getter/setter accessors - If
accessing any attribute for the view object, always use the typesafe getter and setter methods provided by the ViewRowImpl over
the generic getAttribute and setAttribute methods.
Use type-safe bind variable getter/setter accessors - If
accessing bind variable for the view object, always use the typesafe getter and setter methods provided by the ViewObjectImpl
over the generic getAttribute and setAttribute methods.

Check view object SQL performance - Ensure to check the

performance in the database of the underlying view object
queries, especially critical for search forms. If need be seek
guidance from your DBA on creating the appropriate database
indexes for the search criteria.
Consider view object fetch size and access mode - Consider
the view object tuning options for how the view object will be used.
If the view object is used to display a page of 10 records for
example, set the Batch/Fetch size of the view object to 10 + 1 too.
Use createRowSetIterator() for view object programmatic
iteration - if in code you need to programmatically iterate over the
rows provided by a view object, use the createRowSetIterator()
and closeRowSetIterator() methods to manage a separate
RowSetIterator to the default provided by the view object, as the
default may be bound to the user interface
Use view criteria over editing view object where clauses Rather than customizing a view objects where clause which
makes the use of the view object less flexible, use a viewcriteria
instead.

Application Module

Use optimistic locking - Only use optimistic locking (the default)

for web applications, rely on the global setting in the adfconfig.xml file to set this.
Use application module JDBC data source only - Only use
JDBC data sources for application module configurations, do not
use JDBC URLs.
Close callableStatement and preparedStatements - Via the
ApplicationModuleImpl you can retrieve a DBTransactionImpl to
call createCallableStatement or createPreparedStatements.
These both return JDBC objects which require you to close the
statements once complete. This requires a try/catch/final block to
ensure the JDBC connection is closed at the end of its use.
Carefully consider using transactions within overridden ADF
BC code Besides custom methods, generally speaking ADF BC
manages user transactions on the developers behalf. Any usage
of get(DB)Transaction() .commit() or rollback() should be
considered very carefully based on the possible side effects it may
have to open transactions within the AM scope.

Use view object typesafe getters - By default the

AppModuleImpl contains getter methods to return the associated
data model view objects exposed through the application module
editor. Ensure for the relating view objects that you've first
generated the associated ViewObjectImpl class and that the getter
method in the AppModuleImpl returns the typesafe class (e.g
CustomerViewImpl) rather than the parent ViewObjectImpl.

Extend custom exceptions with JboException or subclasses

Generally speaking custom Java exceptions in code should be
avoided and applications should rely on declarative validation
instead. However if an application does write any programmatic
exceptions all custom exceptions written in the ADF BC model
layer should extend JboException or a valid subclass such as
AttrValException that when raised the error is handled gracefully
by the UI layer.

Do not import view or controller classes - The model project

should not import or use classes from the view or controller layers
such as oracle.adf.controller, oracle.adf.view or
org.apache.myfaces.trinidad otherwise they are in violation of the
MVC pattern.
Don't hardcode human readable text - Do not hardcode human
readable text in your ADF Business Component Java code.
Rather store any hardcoded text in a resource bundle and refer to
the key value-pair instead.
Don't introduce HTTP session dependencies into ADF BC,
rely of ADFContext where possible - ADF BC code should
maintain separation and not assume that it is being
called from within the context of an ADF Web Application. The
HTTP session and other UI layer states should not be accessed
directly if you need to use the ADFContext object to access these
instead.

Other General ADF BC Best Practices


Define a jbo.debugoutput run option - Within the

ADF Business Component model project properties
define a new Run/Debug/Profile option entitled
JboDebugOutputConsole where the Launch Setting
Java options is set to -Djbo.debugoutput =console.
Only use this setting for none production systems, it is
not supported in production ADF applications.

Ensure jbo.debugoutput disabled for production Ensure the Java command line Djbo.debugoutput=console option is not enabled for
production ADF systems.

Pages

Use JSPX rather than JSP for 11gR1 applications - If using a

JDeveloper 11gR1 release (which doesnt use Facelets), use
JSPX XML compliant files rather than JSP.
Don't hardcode human readable text - Do not hardcode human
readable text in your pages/fragments. Rather store any
hardcoded text in a resource bundle and refer to the key-value
pair instead.
Avoid inline HTML use - Embedded HTML in JSF pages is often
responsible for breaking the behavior of the ADF Faces Layout
components. Additionally HTML embedded in a JSF page
assumes that the page is indeed being rendered in a conventional
browser and may link the application to a particular subset of
browsers.
Consider command partialSubmit=true - Command
components such as af:commandButton by default when pressed
cause a complete page refresh with the server. If not navigating
between pages, consider using partialSubmit=true which will
result in only components with their partialTriggers set to be
processed.



Avoid long component ID lengths - Avoid long component ID

names as this has a performance hit.
Use rendered=false over visible=false if possible - If you wish
to hide a component use rendered=false over visible=false when
possible as the first results in the server skipping the processing of
the component and sending it over the wire to the browser, while
the 2nd requires the server to process the component, send it
over the wire, the browser to process it then hide it.
Consider that visible=false presents a security risk for
components As components with visible=false are still sent to
the client then hidden, it's easy for hackers to see them in the
browser DOM, manipulate them and if they're submittable
components send them on the next request. Instead either use
rendered=false so the component isn't sent to the browser in the
first place, or back the submittable component by a validation rule
to ensure it can't be submitted when the component is meant to
be hidden.

Never rely on UI side validation - Never solely rely on UI

(browser) validation alone as this can be easily exploited by
hackers. As example a selectOneChoice on the browser provides
an allowable list of values the user can select from, but doesn't
validate that the user hasn't submitted a different value not in the
list. Always back UI side validation up with server side validation
to catch such
workarounds.
Avoid hardcoding EL literals - Avoid hardcoding literals in EL
expressions. Instead place the hardcoding in a managed bean
and refer to it through a boolean accessor instead. For example
the EL expression #{bindings.Status.inputValue == 'Updateable'}
could be replaced with #{editBookings.updateable}.
Avoid repeating EL conditions - If there is a common EL
condition used through out a page consider placing the logic in a
bean boolean accessor instead.

Restrict the use of EL evaluation in code - Evaluating

expression language within (and only within) Java Code can be
error prone and requires a context switch which is inherently slow.
It is preferable to inject EL values as managed properties to a
bean and therefore allow the design time to validate the reference,
or if the values must be determined at the point the code is
evaluated, access the values via Java context objects or other
Java constructs.
Do not prefix EL references to managed beans in request,
session or application scope with the *Scope prefix - Although
it seems to be a good thing to be explicit about the scope that you
expect such an object to be in, doing so will bypass the JSF
creation routines if the bean does not already exist. This is not an
issue for the ADF specific scopes.

JavaScript



Component ID Lookup using

AdfPage.PAGE.findComponentByAbsoluteId() Component lookup in this way, with the complete and absolute ID
of the component in question is the most reliable.
JavaScript strings invoked from the
ExtendedRenderKitService should be minimal - This ensures
that the packets sizes for roundtrips are kept to a minimum and
also ensures that the code that is invoked in this way is subject to
the same scrutiny as other JavaScript used by the system.
Avoid inline page/fragment JavaScript and CSS - Avoid placing
custom JavaScript and CSS in your page and page fragments as
these must be loaded by the browser each time the page/fragment
is accessed. Instead place the JavaScript and CSS respectively in
separate .JS and .CSS files so they can be cached by the
browser.

Avoid inlineStyle and contentStyle CSS added to a page CSS added directly in a page is not compressed, it makes the
page harder to manage and maintain, and increases the page
download size. Use inlineStyle and contentStyle properties for
dynamic color coding only and have the properties referencing a
managed bean property to return the CSS string
Do not use percentage height styles - Heights expressed as a
percentage are not reliable and behavior differs between
browsers. Use absolute values for height if height must be
specified at all

Bindings

Only interact with the business model through the binding

layer - Accessing the underlying business model (e.g. an
application module) via the dataprovider can have multiple
negative effects such as causing memory and connection leaks,
getting the wrong sessions data and breaking the auto PPR
functionality.
Dont reference #{data} - The #{data} expression root is not
guaranteed to provide access to the required data. Success will
depend on the state of the cache. If data is needed in a particular
view / method then create an explicit binding in that scope.

Taskflows

Define a common task flow template - All bounded task flows

can be designed to inherit from a single task flow template that
can be used to provide common functionality like exception
handling and logging.
Use required parameters - Use the required attribute for any
parameter to ensure that the design time and runtime
environments can identify that a particular parameter is optional or
required.
Avoid multiple task flow definitions in one task flow file - for
each task flow XML metadata file restrict the file to one task flow
definition.
Unbounded task flow should only be used for entry points Any view exposed through the adfc-config unbounded Task Flow
is accessible via the URL and open to user manipulation. As such
the actual pages exposed in this way should be restricted.

Use bound methods on task flows rather than invokeAction

executables in the pageDef - Although the use of invokeAction is
allowed it is generally much clearer to use a method activity in a
page flow to explicitly show when code is executed before a page
or fragment is displayed. Also the invokeAction mechanism has no
way to cleanly execute associated code once before the page
renders, using the method activity provides you with the cleanest
interface for implementing this logic.

Beans

Limit custom code in bean accessors - As the JSF lifecycle

may call the getter and setter method in your beans multiple
times, keep the code as small as possible to avoid unnecessary
processing.
Use beans only defined in the task flow for reuse - To make
them reusable, taskflows should not reference managed beans
that weren't defined within the task flow.
Avoid applicationScope and sessionScope variables - Use
applicationScope and sessionScope variables with care as they
create a tight coupling and unintended dependencies within code.
Use unbounded task flow scope pageFlowScope variables
over sessionScope to support multi browser tabs - ADF
supports users opening multiple browser tabs onto your
application. From the user's perspective these are two separate
connections to the application, and as such the state for each
should be separate. Variables stored in sessionScope will be
shared across both browser tabs as only one instance of the
sessionScope bean will be instantiated. Alternatively for
pageFlowScope beans one is instantiated per browser tab.

Define applicationScope and sessionScope beans in adfcconfig.xml - applicationScope and sessionScope managed
beans should be defined in the adfc-config.xml file.
Define unbounded task flow beans in adfc-config.xml - For
any page or page fragment used by an application's unbounded
task flow, any relating bean used by the page should also be
defined in the adfc-config.xml.
Define bounded task flow bean in their own .xml file - For any
page or page fragment used by a bounded task flow, any relating
managed bean used by the page should also be defined in the
bounded task flow's .xml file.
Avoid business logic in bean methods - complex business logic
that manipulates model layer data should not be placed in the
ViewController layer. Consider placing such code in the model
layer instead and making one programmatic call from the bean.
Variables in applicationScope and sessionScope beans - Be
mindful of the variables you store in applicationScope and
sessionScope as they may be stored there for longer than
necessary.

Define all managed beans explicitly in XML rather than

through a mix of XML and annotations - As of JSF2 we can use
annotations in code to define classes as managed beans in the
standard JSF scopes. However, such annotations do not exist for
the ADF specific scopes and mixing both techniques may make it
harder to track down the definition when maintaining the code.
Only create backing beans when actually needed, do not bind
unnecessary component references into backing beans - You
only need a Backing Bean when you actually need somewhere to
hold event handler code. If you have no such code requirements,
dont bother to create the bean.
Utilize common utility class or Backing Bean superclass for
common functions - Utility Methods are often reused within
multiple backing beans and a common provider class can save a
lot of time and coding mistakes

Other General Java Best Practices

Don't use internal ADF packages - Do not use any of the

following ADF packages within your Java code as Oracle does not
guarantee to change the implementation in the future
Don't use System.exit() - Executing System.exit() in a running
ADF application will force the JVM container to quit. Do not use
this.
Don't hardcode human readable text that will be displayed in
the UI - Do not hardcode human readable text in your Java code
or web pages that will be displayed to the user. Rather store any
hardcoded text in a resource bundle and refer to the key valuepair instead.
Ensure catch blocks do not swallow errors - Stubbed try/catch
blocks that do not re-throw or report exceptions by logging may be
concealing bugs. Rather ensure all catch blocks rethrow
exceptions so they percolate through the ADF controller or ADF
model exception handlers.
Include Javadoc - provide an appropriate level of documentation
through out your code.

Don't use System.out.* or System.err.* - Do not use functions

like System.out.println.for debugging and do not leave this
unnecessarily in your code.
Use the ADFLogger- Make use of ADFLogger in your ADF
applications.
Consider user level logging in production Enterprise
manager now supports user level tracing which allows you to
lower the log level for a single session. Traditionally turning on
logging on production systems would result in a flood of
information for all sessions making the reason to write logging in
the first place less than useful. However with the ability to turn
logging on for a specific user this makes the implementation of
logging more useful.
Consider logging granularity - Dont write all log messages at
the same logging level. Consider what information will be of use
for respectively; reporting, error details, live analysis by support
and specialized analysis by the development team.


Always reformat code - All code should be

reformatted before check in to assist readability.

Remove commented out code - Do not leave
commented out code in your source code upon check
in.

web.xml

Consider setting
oracle.adf.view.rich.security.FRAME_BUSTING for production
consider the web.xml
oracle.adf.view.rich.security.FRAME_BUSTING context
parameters options to prevent clickjacking, which occurs when a
malicious web site pulls a page originating from another domain
into a frame and overlays it with a counterfeit page.
Consider oracle.adf.view.rich.versionString.HIDDEN=false for
production set the web.xml
oracle.adf.view.rich.versionString.HIDDEN context parameter to
true to remove the ADF Faces version information on each page
to reduce the size of the page and also remove an attack vector
by unnecessarily publishing the version of ADF.

Deployment to Production Systems Versus

Development/Testing Systems
org.apache.myfaces.trinidad.CHECK_FILE_MODIFICATION=false
If this parameter is true, there will be an automatic check of the
modification date of your JSPs, and saved state will be discarded
when JSP's change. It will also automatically check if your skinning
css files have changed without you having to restart the server.
This makes development easier, but adds overhead. For this
reason this parameter should be set to false when your application
is deployed in environments other than Development instances.

weblogic.xml


url.rewriting.enabled should be false - If enabled,

the feature would cause security vulnerability because
the session identifier would be exposed in certain
scenarios.

Deployment


Use ojdeploy for Packaging - For applications

leveraging ADF Security or MDS, this is an important
step to ensure that all of the correct metadata is
included in the archive.

Ensure connection information is not deployed with
the application - Connections should be configured by
the consumer of the application usually though the
Java EE data source mechanism.

Packaged EAR file should include information
bearing Manifiest file

Use ADF Libraries appropriately for reusable
components - ADF Libraries provide the best
mechanism to ensure the discoverability of metadata
based artifacts within a standard JAR format.

References
1) Oracle ADF Technical overview:
http://www.oracle.com/technetwork/developertools/adf/adf-11-overview-1-129504.pdf
2) Oracle ADF Web Developers Guide 11.1.2.1.0
http://docs.oracle.com/cd/E24382_01/web.1112/e1618
1.pdf
3) Oracle ADF Fusion Developer's Guide 11.1.2.1.0
http://docs.oracle.com/cd/E29049_01/web.1112/e1618
2.pdf