Jump to: navigation, search

Difference between revisions of "FAQ What is the difference between a command and an action?"

m (References)
 
(8 intermediate revisions by 2 users not shown)
Line 1: Line 1:
Commands and actions are two aspects of the same thing.  You can think of commands as the declarative part—specified  in the plug-in manifest—and actions as the  programmatic part.  Commands are used mainly to associate customizable key bindings with actions.  If you look in the '''Workbench > Keys''' preference page, you will see a list of all commands known to the platform, including what context and configuration they belong to. Key bindings are hooked to commands, and then commands are hooked to actions.  This extra level of indirection allows for added flexibility in the implementation. The user can change key bindings for a command without the associated actions knowing about it, and the action for a command can be dynamically changed for different circumstances without affecting the key bindings.
+
Since you have come this far, you probably already understand that Actions and Commands basically do the same thing: They cause a certain piece of code to be executed. They are triggered, mainly, from artifacts within the user interface. These artifacts can be an icon in a (tool)bar, a menu item or a certain key combination.  
  
Note that although the Eclipse APIs contain both commands and actions, the UI very clearly uses only the term <i>command</i> when  referring to them. The term <i>action</i> is avoided in the user interface. There&#146;s no point in confusing users with two names for the same thing!
+
The action framework is proven, tightly integrated and fairly easy to program. So, why change?
 +
 
 +
==Actions==
 +
The main concern with Actions is that the manifestation and the code is all stored in the Action. Although there is some separation in Action Delegates, they are still connected to the underlying action. Selection events are passed to Actions so that they can change their enabled state (programmatically) based on the current selection. This is not very elegant. Also to place an action on a certain workbench part you have to use several extension points :
 +
 
 +
* org.eclipse.ui.viewActions
 +
* org.eclipse.ui.popupMenus
 +
* org.eclipse.ui.editorActions
 +
 
 +
==Commands==
 +
Commands pretty much solve all these issues. The basic idea is that the Command is just the abstract idea of some code to be executed. The actual handling of the code is done by, well, handlers. Handlers are activated by a certain state of the workbench. This state is queried by the platform core expressions.  This means that we only need one global ''Save'' command which behaves differently based on which handler is currently active. Although this specific Command could also be retargeted by a global action, this still has to be done programmatically and not declaratively. To place a Command on a certain workbench part (including the trim area) you have to use only one extension point:
 +
 
 +
* org.eclipse.ui.menus
 +
 
 +
Besides this, Handlers can be activated by a powerful expression syntax in the manifest. This means less code and more declarations which will lead to a smoother running workbench or RCP application.
 +
 
 +
If you look in the '''General &gt; Keys''' preference page, you will see a list of all commands known to the platform, including what context (the "When" column) and configuration (the "Scheme" drop-down list?) they belong to. Key bindings are hooked to commands, and then commands are hooked to handlers. This extra level of indirection allows for added flexibility in the implementation. The user can change key bindings for a command without the associated handlers knowing about it, and the handler for a command can be dynamically changed in different circumstances.
 +
 
 +
==References==
 +
* [http://wiki.eclipse.org/Platform_Command_Framework Platform Command Framework]
 +
* [http://wiki.eclipse.org/Platform_Expression_Framework Platform Expression Framework]
 +
* [http://wiki.eclipse.org/Command_Core_Expressions Core Expressions]
 +
* [http://wiki.eclipse.org/Menu_Contributions Menu Contributions]
 +
* [http://wiki.eclipse.org/Menus_Extension_Mapping Menu Extension Mapping]
 +
* [http://help.eclipse.org/help33/index.jsp?topic=/org.eclipse.platform.doc.isv/guide/workbench.htm Eclipse Workbench Guide]
 +
* [http://blog.rcp-company.com/2007/06/working-with-menus-extension-point.html Working with mnu extension point]
 +
* [https://bugs.eclipse.org/bugs/show_bug.cgi?id=223445 bug 223445 discusses references to an upcoming article]
  
 
== See Also: ==
 
== See Also: ==
*[[FAQ How do I make key bindings work in an RCP application?]]
+
* [[FAQ How do I make key bindings work in an RCP application?]]
*[[FAQ How do I associate an action with a command?]]
+
* [[FAQ How do I associate an action with a command?]]
  
 
{{Template:FAQ_Tagline}}
 
{{Template:FAQ_Tagline}}

Latest revision as of 05:13, 12 November 2010

Since you have come this far, you probably already understand that Actions and Commands basically do the same thing: They cause a certain piece of code to be executed. They are triggered, mainly, from artifacts within the user interface. These artifacts can be an icon in a (tool)bar, a menu item or a certain key combination.

The action framework is proven, tightly integrated and fairly easy to program. So, why change?

Actions

The main concern with Actions is that the manifestation and the code is all stored in the Action. Although there is some separation in Action Delegates, they are still connected to the underlying action. Selection events are passed to Actions so that they can change their enabled state (programmatically) based on the current selection. This is not very elegant. Also to place an action on a certain workbench part you have to use several extension points :

  • org.eclipse.ui.viewActions
  • org.eclipse.ui.popupMenus
  • org.eclipse.ui.editorActions

Commands

Commands pretty much solve all these issues. The basic idea is that the Command is just the abstract idea of some code to be executed. The actual handling of the code is done by, well, handlers. Handlers are activated by a certain state of the workbench. This state is queried by the platform core expressions. This means that we only need one global Save command which behaves differently based on which handler is currently active. Although this specific Command could also be retargeted by a global action, this still has to be done programmatically and not declaratively. To place a Command on a certain workbench part (including the trim area) you have to use only one extension point:

  • org.eclipse.ui.menus

Besides this, Handlers can be activated by a powerful expression syntax in the manifest. This means less code and more declarations which will lead to a smoother running workbench or RCP application.

If you look in the General > Keys preference page, you will see a list of all commands known to the platform, including what context (the "When" column) and configuration (the "Scheme" drop-down list?) they belong to. Key bindings are hooked to commands, and then commands are hooked to handlers. This extra level of indirection allows for added flexibility in the implementation. The user can change key bindings for a command without the associated handlers knowing about it, and the handler for a command can be dynamically changed in different circumstances.

References

See Also:


This FAQ was originally published in Official Eclipse 3.0 FAQs. Copyright 2004, Pearson Education, Inc. All rights reserved. This text is made available here under the terms of the Eclipse Public License v1.0.