Understanding the application context system in profiling views

Profiling views can display profiling data from any programming language or data collection context. This is called the "application context" system for profiling views.

For example, the Java term "method," the C term "function," and the Fortran term "procedure" all refer to the same idea. The application context system allows each profiling agent to indicate the name of the programming language or other domain (that is, the "application context" name) that the data comes from. Using this information, the profiling views will display the data using the proper terms and formatting rules.

The profiling views will display the proper terminology for a data set's application context only if your workbench has the proper "context formatter" plug-in installed. "Context formatters" are plug-ins that supply sets of words and formatting methods to the profiling views. If you do not have a context formatter for the data's application context, the views will display the data using the default Java context formatter. Typically, a data collector and the corresponding context formatters are packaged together.

A single profiling data set might contain items from more than one application context. For example, a single execution trace might include both Java methods and C functions. When the profiling views display a mixed-content data set, they also display an indicator that shows the application context of the selected item and a drop-down list that lets you choose which application context the view should use overall. The application context choice controls the terms used in column headers for tables and the names of toolbar buttons, among other things.

In addition, toolbar drop-down menu has an "Application contexts" submenu that choose the context that should be used for the current profiling view.

Both the drop-down menu and the "Application contexts" submenu list all the application contexts that appear in the current data set, and indicate which one is currently being used by the view for table column headers and other items.

As you read the documentation about profiling features and compare it to the profiling views you see, you should remember that the documentation is based on the default Java application context. If you are viewing data from a different application context, the views will work as documented, but the words you see and some of the syntax and formatting might be different.