Oracle ADF Best Practices, Mistakes and Worst Practices

In this post I am putting some practices and that we should follow while using Oracle Application Development Framework for development

• Call database (PL/SQL) function and procedure in model layer and then access in managed bean using page level bindings
  ADF Basics: Call PL/SQL Stored function in ADF Application
• Use Exception Handler in Taskflow where required
  Global Exception Handler for ADF Task Flow Method Calls
• Some ADF Component use animation for rendering like table, tree, popup. Consider animation-enabled false for better performance
  This can be controlled through ADF Skin
• Ensure that each EnityObject/ ViewObject must have a primary key







• Configure resource bundle before creating business components
  Using external XML file as list resource bundle in ADF Application
• Avoid code duplication, Use utility classes instead - ADFUtil, JSFUtil etc
  ADF and JSF Utils Clasees
• Avoid using editable tables, Use forms to edit data of tables
• Avoid editable primary key attribute
• Close iterators after use
• Use ADF Logger instead of System.out.println
• Choose bean scope carefully , lots of problem occurs due to wrong bean scope
  Memory Scope For ADF Managed Beans-As per Fusion Developer guide
• Don't use getDBTransaction().postChanges unless you have a commit operation in same request
• Check for null values instead of using try/catch block everywhere
• Avoid circular dependency of list of values
• Use createRowSetIterator() for view object programmatic iteration instead of using AllRowsInRange
• Always use readOnly viewObject for view based or reference data, This helps in performance as it avoids EntityCache
• Write business logic in model level implementation classes as AMImpl/VOImpl and call those methods in bean using binding layer
  ADF Basics: How to invoke model layer methods from managed bean (Best Practice to write business logic in ADF)
• Implement bean serialization for applicationScope, sessionScope, pageFlowScope beans 
• ADF Faces requires only one root component per fragment
• Ensure uniqueness of taskflow name, managed bean name and application module name in entire application
• Always check for null values while calculating transient attributes default values
• Don't write bulky and unnecessary code in accessors of components or attributes as accessors are called multiple times so it'll affect performance
• Catch Jbo Exceptions while calling pl/sql functions/procedure and show as a message on page instead of showing ADF/DB generated messages
• To apply validation and autoSuggest like feature in search form use custom search instead of af:query component
• Ensure that component bindings should be in a request scoped bean not greater than that
• Instead of using setWhereClause() use query based bind variable to improve performance
• Use basic tuning in ViewObjects to fetch rows from DB to improve performance
• Don't set immediate true for editable input components
• Don't use cached application module reference in bean instead of caching, look for AM when required as it is a pooled resource that's why it is never guaranteed that you will get same AM instance over a period of time. You'll end up getting lots of Null Pointer Exceptions
• Don't use RowId as primary key in entity object , To avoid this define primary key in DB table itself
• Never use SELECT * FROM in viewObject instead specify required column names
• Avoid ViewObject expert mode for entity based viewObject. To add where clause , order by clause use viewObject or view object based bind variable