HP OpenVMS Systems Documentation

Content starts here
Common Desktop Environment: Advanced User's and System Administrator's Guide

10 Creating Actions Manually


Contents of Chapter:
Reasons You Must Create an Action Manually
COMMAND Actions
MAP Actions
TT_MSG (ToolTalk Message) Actions
Creating an Action Manually: General Steps
Configuration Files for Actions
To Create an Action Manually
Example of Creating a COMMAND Action
Example of Creating a MAP Action
To Reload the Actions/Data Types Database
Creating an Action File (Icon) for an Action
To Create an Action File (Action Icon)
Specifying the Icon Image Used by an Action
To Modify an Existing Action Definition
Precedence in Action Definitions
Building the Execution String for a COMMAND Action
General Features of Execution Strings
Creating an Action that Uses No Arguments
Creating an Action that Accepts a Dropped File
Creating an Action that Prompts for a File Argument
Creating an Action that Accepts a Dropped File or Prompts for One
Creating an Action that Prompts for a Non-File Argument
Interpreting a File Argument as a String
Providing Shell Capabilities in an Action
Creating COMMAND Actions for Multiple File Arguments
Windowing Support and Terminal Emulators for COMMAND Actions
Specifying the Window Support for the Action
Specifying Command-Line Options for the Terminal Emulator
Specifying a Different Default Terminal Emulator
Restricting Actions to Certain Arguments
Restricting an Action to a Specified Data Type
Restricting an Action Based on the Number of Arguments
To Provide Different Double-Click and Drop Behavior
Restricting an Action Based on the Mode of the Argument
Creating Actions that Run Applications on Remote Systems
Creating an Action that Runs a Remote Application
Using Variables in Action and Data Type Definitions
Using String Variables in an Action
To Define a String Variable
To Reference a String Variable
Using Environment Variables in Actions and Data Types
Invoking Actions from a Command Line
Syntax of dtaction
Creating an Action that Runs Another Action
Creating an Action that Runs as a Different User
Creating Localized Actions
Locations for Localized Actions
To Localize an Existing Action
Creating Actions for ToolTalk Applications
There are two ways to create actions:

  • Using the Create Action desktop application
  • Manually creating an action definition
Creating an action manually requires you to edit a database file. This chapter describes how to manually create action definitions.

See Also

Reasons You Must Create an Action Manually

There are three basic types of actions:

  • COMMAND
  • MAP
  • TT_MSG
The Create Action tool is designed to create certain types of COMMAND and MAP actions. All TT_MSG actions must be created manually.

For more information, see "Limitations of Create Action".

COMMAND Actions

A command action executes a command that starts an application or utility, runs a shell script, or executes an operating system command. The definition of the action includes the command to be executed (the EXEC_STRING).

The Create Action tool can be used to create the most common types of command actions. However, there may be some situations where you must create the action manually; for example, you must create a COMMAND action manually if the action specifies:

  • Multiple-file arguments with a different prompt for each argument.

  • Action invocation--the ability of actions to invoke other actions.

  • Argument-count dependent behavior--the ability to create an action that has very different behaviors for different numbers of file arguments.

  • A remote execution host--the ability to run an application on a system other than the one containing the action definition.

  • Change of user--the ability to run the action as a different user (for example, to prompt for the root password and then run as root).

MAP Actions

A map action is an action that is ``mapped'' to another action rather than directly specifying a command or ToolTalk message

Mapping provides the ability to specify alternative names for actions. For example, a built-in command action named IconEditor starts Icon Editor. The database also includes an Open action, restricted in the definition to bitmap and pixmap files (by the ARG_TYPE field), that is mapped to the IconEditor action. This lets the user start Icon Editor by selecting a bitmap or pixmap file in File Manager and then choosing Open from the Selected menu.

Create Action provides limited mapping for Open and Print actions. All other map actions must be created manually.

TT_MSG (ToolTalk Message) Actions

TT_MSG actions send a ToolTalk message. All TT_MSG actions must be created manually.


Creating an Action Manually: General Steps

This section explains how to create a configuration file for an action definition.

Configuration Files for Actions

Configuration files containing action definitions must meet these requirements:

For information on modifying the actions/data types search path, see "Setting the Value of a Search Path".

To Create an Action Manually

  1. Open an existing database file or create a new one. See the previous section, "Configuration Files for Actions".

  2. Create the action definition using the syntax:

    ACTION action_name
    {
       TYPE     action_type
       action_field
       ...
    }

    where:

    action_name
    Name used to run the action.

    action_type
    COMMAND (default), MAP, or TT_MSG.

    action_field
    One of the required or optional fields for this type of action. All fields consist of a keyword and a value.

    Many of the action fields are covered in this chapter. For more information, see the dtactionfile(4) man page.

  3. Save the file.

  4. If you want the action icon to have a unique image, create the icons for the action. The default location for icons is:

    • Personal icons: HomeDMcClurg irectory/.dt/icons

    • System-wide icons: /etc/dt/appconfig/icons/language. The default language is C.

    For more information, see "Specifying the Icon Image Used by an Action".

  5. Double-click Reload Actions in the Desktop_Tools application group.

  6. Create an action file for the action. The action file creates an icon in File Manager or Application Manager that represents the action. (If the action is written to start an application, the icon is called an application icon.)

    To create the action file, create an executable file with the same name as action_name. You can put the file in any directory to which you have write permission. You can create as many action files as you like.

Example of Creating a COMMAND Action

The following steps create a personal action that starts a fax application on remote system AppServerA. The command for starting the fax application is:

/usr/fax/bin/faxcompose [filename]

  1. Create the file HomeDirectory/.dt/types/Fax.dt.

  2. Put the following action definition into the file:

    ACTION FaxComposer
    {
       TYPE           COMMAND
       ICON           fax
       WINDOW_TYPE    NO_STDIO
       EXEC_STRING    /usr/fax/bin/faxcompose -c %Arg_1%
       EXEC_HOST      AppServerA
       DESCRIPTION    Runs the fax composer
    }

    The WINDOW_TYPE and EXEC_STRING fields describe the behavior of the action.

    WINDOW_TYPE
    The NO_STDIO keyword specifies that the action does not have to run in a terminal emulator window.

    See "Specifying the Window Support for the Action".

    EXEC_STRING
    The syntax %Arg_1% accepts a dropped file. If the action icon is double-clicked, the action opens an empty fax composer window.

    See "Building the Execution String for a COMMAND Action".

  3. Save the file.

  4. Use Icon Editor to create the following icon image files in the HomeDirectory/.dt/icons directory:

    • fax.m.pm, size 32 by 32 pixels

    • fax.t.pm, size 16 by 16 pixels

  5. Double-click Reload Actions in the Desktop_Tools application group.

  6. Create an executable file named FaxComposer in a directory to which you have write permission (for example, your home directory).

Example of Creating a MAP Action

Suppose most of the files you fax are created with Text Editor and are of data type TEXTFILE (files named *.txt).

These steps add a ``Fax'' menu item to the data type's Selected menu.

  1. Open the file HomeDirectory/.dt/types/Fax.dt that was created in the previous example.

  2. Add this map action definition to the file:

    ACTION Fax
    {
       ARG_TYPE       TEXTFILE
       TYPE           MAP
       MAP_ACTION     FaxComposer
    } 

  3. Save the file.

  4. Copy the data attributes definition for TEXTFILE from /usr/dt/appconfig/types/language/dtpad.dt to a new file HomeDirectory/.dt/types/textfile.dt. Add the Fax action to the ACTIONS field.

    DATA_ATTRIBUTES TEXTFILE
    {
       ACTIONS        Open,Print,Fax
       ICON           Dtpenpd
       ...
    }

  5. Save the file.

  6. Open Application Manager and double-click Reload Actions in the Desktop_Tools application group.

To Reload the Actions/Data Types Database

In order for new or edited action definitions to take effect, the desktop must reread the database.

Open the Desktop_Tools application group and double-click Reload Actions.

Or, execute the command:

dtaction ReloadActions

ReloadActions is the name of the action whose icon is labeled ``Reload Actions.''

The actions database is also reread when the user:

  • Logs in

  • Restarts the Workspace Manager

  • Saves an action in the Create Action window by choosing Save from the File menu

Creating an Action File (Icon) for an Action

An action file is a file created to provide a visual representation of the action in File Manager or Application Manager.

Figure 10-1 Action files (also called action icons or application icons) in Application Manager

Since an action file's icon represents an action, it is sometimes called an action icon. If the underlying action starts an application, the action file icon is called an application icon.

Double-clicking the action icon runs the action. The action icon may also be a drop zone.

To Create an Action File (Action Icon)

Create an executable file with the same name as the action name. The content of the file does not matter.

For example, if the action definition is:

ACTION  MyFavoriteApp
{
   EXEC_STRING     Mfa -file %Arg_1%
   DESCRIPTION     Runs MyFavoriteApp
   ICON            Mfapp
}

then the action file would be an executable file named MyFavoriteApp. In File Manager and Application Manager, the MyFavoriteApp file would use the icon image Mfapp.size.type. Double-clicking MyFavoriteApp's icon would run the action's execution string, and the icon's On Item help would be the contents of the DESCRIPTION field (Runs MyFavoriteApp).

Action Labels

If the action definition includes the LABEL field, the action file will be labeled in File Manager and Application Manager with the contents of this field rather than the file name (action_name). For example, if the action definition includes:

ACTION  MyFavoriteApp
{
   LABEL      Favorite Application
   ...
}

then the action icon will be labeled ``Favorite Application.''

Specifying the Icon Image Used by an Action

Use the ICON field to specify the icon used in File Manager and Application Manager for the action icons created for the action.

If you do not specify an icon, the system uses the default action icon image files /usr/dt/appconfig/icons/language/Dtactn.*.

Figure 10-2 Default action icon image

The default action icon can be changed using the resource:

*actionIcon:    icon_file_name

where icon_file_name can be a base name or absolute path.

The value of the ICON field can be:

  • A base file name.

    The base file name is the name of the file containing the icon image minus the file-name suffixes for size (m and t) and image type (bm and pm). For example, if files are named GameIcon.m.pm and GameIcon.t.pm, use GameIcon.

    If you use the base file name, the icon files must be placed in a directory on the icon search path:

  • Personal icons: HomeDirectory/.dt/icons

  • System-wide icons: /etc/dt/appconfig/icons/language

  • An absolute path to the icon file, including the full file name.

    You should use the absolute path only if the icon file is not located on the icon search path. For example, if icon file GameIcon.m.pm is placed in the directory /doc/projects, which is not on the icon search path, the value of the ICON field would be /doc/projects/GameIcon.m.pm.

Table 10-1 lists icon sizes you should create and the corresponding file names.



To Modify an Existing Action Definition

You can modify any of the actions available on your system, including built-in actions.


Note: Use caution when modifying the built-in action database. The built-in actions are designed to work well with the desktop applications.

  1. Locate the definition of the action you want to modify.

    The default locations for action definitions are:

    • Built-in actions: /usr/dt/appconfig/types/language

    • System-wide actions: /etc/dt/appconfig/types/language

    • Personal actions: HomeDirectory/.dt/types

    Your system might include additional locations. To see a list of the locations your system uses for actions, execute the command:

    dtsearchpath -v

    Your system uses the directories listed under DTDATABASESEARCHPATH.

  2. If necessary, copy the text of the action definition to a new or existing file in one of these directories:

    • System-wide actions: /etc/dt/appconfig/types/language

    • Personal actions: HomeDirectory/.dt/types

    You must copy built-in actions, since you should not edit files in the /usr/dt/appconfig/types/language directory.

  3. Edit the action definition. When you are done, save the file.

  4. Double-click Reload Actions in the Desktop_Tools application group.

Precedence in Action Definitions

When the user invokes an action, the system searches the database for a matching action name. When more than one action exists with that name, the system uses precedence rules to decide which one to use.

  • If no other precedence rules apply, the precedence is based on the location of the definition. The following list is ordered from higher to lower precedence:

    • Personal actions (HomeDirectory/.dt/types)

    • System-wide local actions (/etc/dt/appconfig/types/language)

    • System-wide remote actions (hostname:/etc/dt/appconfig/types/language). The remote hosts searched are those listed in the application search path.

    • Built-in actions (/usr/dt/appconfig/types/language)

  • Within a given directory, the *.dt files are read in alphabetical order.

  • Actions restricted by ARG_CLASS, ARG_TYPE, ARG_MODE, or ARG_COUNT have precedence over unrestricted actions. (The default for these four fields is *.)

    Where more than one restriction applies, the precedence order from high to low is:

    • ARG_CLASS

    • ARG_TYPE

    • ARG_MODE

    • ARG_COUNT

    Where more than one restricted ARG_COUNT exists, the precedence order from high to low is:

    • Specific integer value n

    • <n

    • >n

    • *

For example, consider the following portions of action definitions:

ACTION EditGraphics
# EditGraphics-1
{
   ARG_TYPE       XWD
   ...
}

ACTION EditGraphics # EditGraphics-2 { ARG_COUNT 0 ... }

ACTION EditGraphics # EditGraphics-3 { ARG_TYPE * ... }

Double-clicking the EditGraphics action icon starts EditGraphics-2 because no argument is provided and ARG_COUNT 0 has precedence. When an XWD-type file argument is provided, EditGraphics-1 is used because it specified the XWD ARG_TYPE. EditGraphics-3 is used for all other file arguments.


Building the Execution String for a COMMAND Action

The minimum requirements for a COMMAND action are two fields--ACTION and EXEC_STRING.

ACTION action_name
{
    EXEC_STRING execution_string
}

The execution string is the most important part of a COMMAND action definition. It uses syntax similar to the command line you would execute in a Terminal window but includes additional syntax for handling file and string arguments.

General Features of Execution Strings

Execution strings may include:

  • File and non-file arguments
  • Shell syntax
  • Absolute paths or names of executables

Action Arguments

An argument is information required by a command or application for it to run properly. For example, consider the command line you could use to open a file in Text Editor:

dtpad filename

In this command filename is a file argument of the dtpad command.

Actions, like applications and commands, can have arguments. There are two types of data that a COMMAND action can use:

  • Files
  • String data

Using Shells in Execution Strings

The execution string is executed directly, rather than through a shell. However, you can explicitly invoke a shell in the execution string.

For example:

EXEC_STRING            \
      /bin/sh -c \
      'tar -tvf %(File)Arg_1% 2>&1 | \${PAGER:-more};\
      echo "\\n*** Select Close from the Window menu to close ***"'

Name or Absolute Path of the Executable

If your application is located in a directory listed in the PATH variable, you can use the simple executable name. If the application is elsewhere, you must use the absolute path to the executable file.

Creating an Action that Uses No Arguments

Use the same syntax for the EXEC_STRING that you would use to start the application from a command line.

Examples
  • This execution string is part of an action that starts the X client xcutsel.

    EXEC_STRING xcutsel

  • This execution string starts the client xclock with a digital clock. The command line includes a command-line option but requires no arguments.

    EXEC_STRING xclock -digital

Creating an Action that Accepts a Dropped File

Use this syntax for the file argument:

%Arg_n%

or

%(File)Arg_n%

(File) is optional, since arguments supplied to Arg_n are assumed (by default) to be files. (See "Interpreting a File Argument as a String" for use of the %(String)Arg_n% syntax.)

This syntax lets the user drop a data file object on the action icon to start the action with that file argument. It substitutes the nth argument into the command line. The file can be a local or remote file.

Examples
  • This execution string executes wc -w using a dropped file as the
    -load parameter.

    EXEC_STRING wc -w %Arg_1%

  • This example shows a portion of a definition for an action that works only with directory arguments. When a directory is dropped on the action icon, the action displays a list of all the files in the directory with read-write permission.

    ACTION  List_Writable_Files
    {
        ARG_TYPE     FOLDER
        EXEC_STRING /bin/sh -c 's -l %Arg_1% | grep rw-'
        ...
    }

Creating an Action that Prompts for a File Argument

Use this syntax for the file argument:

%(File)"prompt"% 

This syntax creates an action that displays a prompt for a file name when the user double-clicks the action icon.

For example, this execution string displays a dialog box that prompts for the file argument of the wc -w command:

EXEC_STRING wc -w %(File)"Count words in file:"%

Creating an Action that Accepts a Dropped File or Prompts for One

Use this syntax for the file argument:

%Arg_n"prompt"%

or

%(File)Arg_n"prompt"%

This syntax produces an action that:

  • Accepts a dropped file as the file argument.

  • Displays a dialog box that prompts for a file name when the user double-clicks the action icon.

For example, this execution string performs lp -oraw on a dropped file. If the action is started by double-clicking the icon, a dialog box appears prompting for the file name.

EXEC_STRING lp -oraw %Arg_1"File to print:"%

Creating an Action that Prompts for a Non-File Argument

Use this syntax for the non-file parameter:

%"prompt"%

or

%(String)"prompt"%

(String) is optional, since quoted text is interpreted, by default, as string data. This syntax displays a dialog box that prompts for non-file data; do not use this syntax when prompting for a file name.

For example, this execution string runs the xwd command and prompts for a value to be added to each pixel:

EXEC_STRING xwd -add %"Add value:"% -out %Arg_1"Filename:"%

Interpreting a File Argument as a String

Use this syntax for the argument:

%(String)Arg_n%

For example, this execution string prints a file with a banner containing the file name, using the command lp -tbanner filename.

EXEC_STRING lp -t%(String)Arg_1% %(File)Arg_1"File to print:"%

Providing Shell Capabilities in an Action

Specify the shell in the execution string:

/bin/sh -c 'command'
/bin/ksh -c 'command'
/bin/csh -c 'command'

Examples
  • This execution string illustrates an action that uses shell piping.

    EXEC_STRING /bin/sh -c 'ps | lp'

  • This is a more complex execution string that requires shell processing and accepts a file argument.

    EXEC_STRING /bin/sh -c 'tbl %Arg_1"Man Page:"% | troff -man'

  • This execution string requires that the argument be a compressed file. The action uncompresses the file and prints it using lp -oraw.

    EXEC_STRING   /bin/sh -c 'cat %Arg_1 "File to print:"% | \
                  uncompress | lp -oraw'

  • This execution string starts a shell script.

    EXEC_STRING /usr/local/bin/StartGnuClient

Creating COMMAND Actions for Multiple File Arguments

There are three ways for actions to handle multiple file arguments:

  • The action can be run repreatedly, once for each argument. When an EXEC_STRING contains a single file argument and multiple file arguments are provided by dropping multiple files on the action icon, the action is run separately for each file argument.

    For example if multiple file arguments are supplied to the following action definition:

    ACTION DisplayScreenImage
    {
       EXEC_STRING      xwud -in %Arg_1%
       ...
    }

    the DisplayScreenImage action is run repeatedly.

  • The action can use two or more non-interchangeable file arguments. For example:

    xsetroot -cursor cursorfile maskfile  

    requires two unique files in a particular order.

  • The action can perform the same command sequentially on each file argument. For example:

    pr file [file ...]  

    will print one or many files in one print job.

Creating an Action for Non-Interchangeable Arguments

Use one of the following syntax conventions:

  • If you want the action to prompt for the file names, use this syntax for each file argument:

    %(File)"prompt"%

    Use a different prompt string for each argument.

    For example, this execution string prompts for two files.

    EXEC_STRING  xsetroot -cursor %(File)"Cursor bitmap:"% \
                 %(File)"Mask bitmap:"%

  • To accept dropped files, use this syntax for each file argument:

     %Arg_n%

    using different values of n for each argument. For example:

    EXEC_STRING  diff %Arg_1% %Arg_2%

Creating an Action with Interchangeable File Arguments

Use one of the following syntax conventions:

  • To create an action that accepts dropped files and issues a command in the form command file1 file2 ..., use this syntax for the file arguments:

    %Args%

  • To create an action that accepts multiple dropped files, or displays a prompt for a single file when double-clicked, use this syntax for the file arguments:

    %Arg_1"prompt"% %Args%

    The action will issue the command in the form: command file1 file2 ....

Examples
  • This execution string creates an action that executes:

    pr file1 file2

    with multiple file arguments.

    EXEC_STRING pr %Args%

  • This execution string creates an action similar to the previous example, except that the action displays a prompt when double-clicked (no file arguments).

    EXEC_STRING  pr %Arg_1"File(s) to print:"% %Args%

Creating an Action for Multiple Dropped Files

To accept multiple dropped file arguments and execute a command line in the form:

command file1 file2 ...

use the syntax:

%Args%

Examples
  • This execution string executes a script named Checkout for multiple files:

    EXEC_STRING  /usr/local/bin/Checkout \
    %Arg_1"Check out what file?"% %Args%

  • This execution string executes lp -oraw with multiple files:

    EXEC_STRING  lp -oraw %Arg_1"File to print:"% %Args%


Windowing Support and Terminal Emulators for COMMAND Actions

There are several ways that COMMAND actions support windows on the desktop.

  • If the application has its own window, the action can be written to provide no additional window support. This option is also used when an action runs a command that requires no direct user input and has no output.

  • If the application must run in a terminal emulator window, the action can be written to open a window and then run the application. There are several terminal options.

Specifying the Window Support for the Action

Use the WINDOW_TYPE field to specify the type of windowing support required by the action.

WINDOW_TYPE
Windowing Support Provided

NO_STDIO
None. Use NO_STDIO if the application has its own window, or if the command has no visible output.

PERM_TERMINAL
Permanent terminal emulator window. The action opens a terminal window that remains open until the user explicitly closes it. The user can enter data into the window. Use with commands that take some input, produce some output, then terminate (for example,
ls directory).

TERMINAL
Temporary terminal emulator window. The action opens a terminal window that closes as soon as the command is completed. Use with full-screen commands (for example, vi).

Specifying Command-Line Options for the Terminal Emulator

Use the TERM_OPTS field in the action definition to specify command-line options for the terminal emulator.

For example, the following action prompts for the execution host:

ACTION OpenTermOnSystemUserChooses
{
   WINDOW_TYPE     PERM_TERMINAL
   EXEC_HOST       %(String)"Remote terminal on:"%
   TERM_OPTS       -title %(String)"Window title:"%
   EXEC_STRING     $SHELL
}

Specifying a Different Default Terminal Emulator

The default terminal emulator used by actions is dtterm. You can change this to another terminal emulator. The default terminal emulator is used when the action does not explicitly specify a terminal emulator to use.

The terminal emulator used by actions must have these command-line options:

  • -title window_title
  • -e command
Two resources determine the default terminal emulator used by actions:

  • The localTerminal resource specifies the terminal emulator used by local applications.

    *localTerminal:    terminal

    For example:

    *localTerminal: xterm

  • The remoteTerminal resource specifies the terminal emulator used by remote applications.

    *remoteTerminal:   host:terminal [,host:terminal...]

    For example:

    *remoteTerminal: sysibm1:/usr/bin/xterm,syshp2:/usr/bin/yterm


Restricting Actions to Certain Arguments

Restricting an action to a particular type of argument refines the action. For example, it is useful to restrict an action that invokes a viewer for PostScript files to only PostScript file arguments; with the restriction, the action will return an error dialog if a non-PostScript file is specified.

You can restrict actions based on:

  • The data type of the file argument.

  • The number of file arguments--for example, no arguments versus one or more arguments. This provides different drop and double-click behavior for the action icon.

  • The read/write mode of the argument.

Restricting an Action to a Specified Data Type

Use the ARG_TYPE field to specify the data types for which the action is valid. Use the data attribute name.

You can enter a list of data types; separate the entries with commas.

For example, the following action definition assumes a Gif data type has been created.

ACTION Open_Gif
{
   TYPE          COMMAND
   LABEL         "Display Gif"
   WINDOW_TYPE   NO_STDIO
   ARG_TYPE      Gif
   ICON          xgif
   DESCRIPTION   Displays gif files
   EXEC_STRING   xgif
}

Restricting an Action Based on the Number of Arguments

Use the ARG_COUNT field to specify the number of arguments the action can accept. Valid values are:

*
(Default) Any number of arguments. Other values have precedence
over *.

n
Any non-negative integer, including 0.

>n
More than n arguments.

<n
Fewer than n arguments.

One use for ARG_COUNT is to provide different action icon behavior, depending on whether the user double-clicks the icon or drops a file on it. See the next section, "To Provide Different Double-Click and Drop Behavior".

To Provide Different Double-Click and Drop Behavior

Use this procedure to create an action that accepts a dropped file but does not prompt for a file when the action icon is double-clicked.

  1. Create an action definition for the double-click functionality.

    Use the ARG_COUNT field to specify 0 arguments. Use a syntax for the EXEC_STRING that does not accept a dropped argument.

  2. Create a second action definition for the drop functionality.

    Use the ARG_COUNT field to specify >0 argument. Use a syntax for the EXEC_STRING that accepts a dropped file.

For example, suppose the following two command lines can be used to start an editor named vedit:

  • To start the editor with no file argument:

    vedit

  • To start the editor with a file argument that is opened as a read-only document:

    vedit -R filename

The following two actions create drop and double-click functionality for an action named Vedit. The first action has precedence when the database is searched for a match, since ARG_COUNT 0 is more specific than the implied ARG_COUNT * of the drop functionality definition.

# Double-click functionality
ACTION Vedit
{
   TYPE            COMMAND
   ARG_COUNT       0
   WINDOW_TYPE     PERM_TERMINAL
   EXEC_STRING     vedit
}

# Drop functionality
ACTION Vedit
{
   TYPE            COMMAND
   WINDOW_TYPE     PERM_TERMINAL
   EXEC_STRING     vedit -R %Arg_1%
}

Restricting an Action Based on the Mode of the Argument

Use the ARG_MODE field to specify the read/write mode of the argument. Valid values are:

*
(Default) Any mode

!w
Non-writable

w
Writable


Creating Actions that Run Applications on Remote Systems

When discussing actions and remote execution, there are two terms that are used frequently:

database host
The system containing the action definition

execution host
The system where the executable runs

In most situations, actions and their applications are located on the same system; since the default execution host for an action is the database host, no special syntax is required.

However, when the execution host is different from the database host, the action definition must specify where the execution string should be run.

The ability to locate actions and applications on different systems is part of the client/server architecture of the desktop. For a more in-depth discussion of networked applications, see "Administering Application Services".

Creating an Action that Runs a Remote Application

Use the EXEC_HOST field in the action definition to specify the location of the application.

Valid values for EXEC_HOST are:

%DatabaseHost%
The host where the action is defined.

%LocalHost%
The host where the action is invoked (the session server).

%DisplayHost%
The host running the X server (not allowed for X terminals).

%SessionHost%
The host where the controlling Login Manager is running.

hostname
The named host. Use this value for environments in which the action should always be invoked on one particular host.

%"prompt"%
Prompts the user for the host name each time the action is invoked.

The default value is %DatabaseHost%, %LocalHost%. Thus, when the EXEC_HOST field is omitted, the action first attempts to run the command on the host containing the action definition. If this fails, the action attempts to run the command on the session server.

Examples
  • This field specifies host ddsyd:

    EXEC_HOST  ddsyd

  • The field prompts for a host name:

    EXEC_HOST  %"Host containing application:"%

  • This field specifies that the action will attempt to run the application on the host containing the action definition. If this fails, the action will attempt to run the application on host ddsyd.

    EXEC_HOST  %DatabaseHost%, ddsyd


Using Variables in Action and Data Type Definitions

You can include string variables and environment variables in action and data type definition files.

Using String Variables in an Action

A string variable definition remains in effect from the location of the definition to the end of the file. There are no global string variables for the database.

If a string variable and environment variable have the same name, the string variable has precedence.

To Define a String Variable

Use the syntax:

set variable_name=value

Variable names can contain any alphanumeric characters and underscore (_). Each variable definition must be on a separate line.

For example:

set Remote_Application_Server=sysapp
set Remote_File_Server=sysdata

To Reference a String Variable

Use the syntax:

$[{]variable_name[}]

For example:

EXEC-HOST    $Remote_Application_Server
CWD         /net/${Remote_File_Server}/doc/project

Using Environment Variables in Actions and Data Types

Reference an environment variable using the syntax:

$[{]variable[}]. 

The variable is expanded (replaced by its value) when the database is loaded. If a string variable and environment variable have the same name, the string variable has precedence.

For example, this execution string prints a file with a banner containing the login name.

EXEC-STRING lp -t$LOGNAME %(File)Arg_1%


Invoking Actions from a Command Line

The desktop provides the dtaction command for running actions from a command line. You can use dtaction to run actions from:

  • Scripts
  • Other actions
  • A terminal emulator command line

Syntax of dtaction

 dtaction [-user user_name] [-execHost hostname] action_name [argument [argument]... ]

-user user_name
Provides the ability to run the action as a different user. If dtaction is invoked by a user other than user_name, a prompt is displayed for the password.

-execHost hostname
For COMMAND actions only; specifies the host on which the command will be run.

argument
Arguments to the action; usually file arguments.

The dtaction client has additional command-line options. For more information, see the dtaction(1) man page.

Creating an Action that Runs Another Action

Use dtaction in the EXEC_STRING of the action.

For example, the following action uses a built-in action named Spell (the action is labeled ``Check Spelling'' in Application Manager). The new action runs Text Editor and the Spell action, displaying the spelling errors in a separate terminal emulator window.

ACTION EditAndSpell
{
   WINDOW_TYPE    NO_STDIO
   EXEC_STRING    /bin/sh -c 'dtaction Spell \
                  %Arg_1"File:"%; dtpad %Arg_1%'
}

Creating an Action that Runs as a Different User

Use the following syntax in the EXEC_STRING:

EXEC_STRING   dtaction -user user_name action_name [file_argument]

The new user (user_name) must have display access to the system through one of the following mechanisms:

  • Read permission on the login user's .Xauthority file
  • Or, xhost permission
For example, the following two actions provide the ability to become root and edit an app-defaults file.

ACTION AppDefaults
{
   WINDOW_TYPE    NO_STDIO
   EXEC_STRING    /usr/dt/bin/dtaction -user root \
                  EditAppDefaults %Arg_1"File:"%
}
ACTION EditAppDefaults
{
   WINDOW_TYPE    TERMINAL
   EXEC_STRING    /bin/sh -c 'chmod +w %Arg_1%; \
                  vi %Arg_1%; chmod -w %Arg_1%'
}


Creating Localized Actions

The search path for data types includes language-dependent locations. The desktop uses the value of LANG to determine the locations searched for data type definitions.

Locations for Localized Actions

Localized action definitions must be placed in the proper language-dependent directories along the actions search path.

The default search path is:

  • Personal actions: HomeDirectory/.dt/types
  • System-wide actions: /etc/dt/appconfig/types/language
  • Built-in actions: /usr/dt/appconfig/types/language

To Localize an Existing Action

  1. Create a file in the appropriate language-dependent directory (for example, in /etc/dt/appconfig/types/japanese) .

  2. Copy the action definition to the language-dependent configuration file.

    For example, you might copy an action definition from

    app_root/dt/appconfig/types/C/file.dt

    to

    app_root/dt/appconfig/types/japanese/newfile.dt

  3. Add a LABEL field or modify the existing LABEL field.

    LABEL			string

    Application Manager and File Manager use the label string to identify the action's icon.

  4. Localize any of the following fields in the action definition:
    • For localized icons: ICON
    • For localized On Item help: DESCRIPTION
    • For localized prompts: any quoted text in the EXEC_STRING

Creating Actions for ToolTalk Applications


Note: The following information applies only to applications that support ToolTalk messaging.

Use the action type TT_MSG to create an action that sends a ToolTalk message.

ACTION action_name
{
   TYPE    TT_MSG
   ...
}

addressing and disposition Fields

  • The ToolTalk addressing field is always set to TT_PROCEDURE.

  • The ToolTalk disposition field defaults to the specification in the static message pattern.

Unsupported Messages

The following are not supported by TT_MSG-type actions:

  • ToolTalk object-oriented messages
  • Context arguments in messages

Keywords for TT_MSG Actions

Keyword
Use

TT_CLASS
Defines the value of the ToolTalk class message field

TT_SCOPE
Defines the value of the ToolTalk scope message field

TT_OPERATION
Defines the value of the ToolTalk operation message field.

TT_FILE
Defines the value of the ToolTalk file message field

TT_ARGn_MODE
Defines the value of the ToolTalk mode attribute for the nth message argument

TT_ARGn_VTYPE
Defines the value of the ToolTalk vtype attribute of the nth message argument

TT_ARGn_VALUE
Defines the value of the nth message argument