Writing a Stored Process

Setting Result Capabilities

A stored process is a SAS program that can produce any type of output that a valid SAS program can produce. Output could include data sets, external files, e-mail messages, SAS catalogs, packages, and many other objects. In some cases, the output (or a result) is delivered to the client application that is executing the stored process. In other cases, the output is generated only on the server.

When you register the stored process, you can specify what type of output the stored process can produce. You can specify Stream, Package, both output types, or neither output type.

When you run the stored process, the client application chooses the type of output that it prefers. For example, when SAS Web Report Studio runs a stored process, package output is produced. There are four types of client output:

The simplest type of output, or result type, is none. The client receives no output from the stored process. The stored process is still able to create or update data sets, external files, or other objects, but this output remains on the server. This result type is indicated when the input parameter _RESULT is set to STATUS because only the program status is returned to the client.
Streaming output delivers a data stream, such as an HTML page or XML document, to the client. This result type is indicated when _RESULT is set to STREAM . The data stream can be textual or binary data and is visible to the stored process program as the _WEBOUT fileref. Any data that is written to the _WEBOUT fileref is streamed back to the client application. Streaming output is supported only on the stored process server. Stored processes that are executing on a workspace server cannot use streaming output.
Package output can be either transient, meaning that the output is returned only to the client and exists only in the current session, or permanent, meaning that the package is stored or published somewhere and can be accessed even after the session ends.
- Transient package output returns a temporary package to the client. The package can contain multiple entries, including SAS data sets, HTML files, image files, or any other text or binary files. The package exists only as long as the client is connected to the server. This result type is a convenient way to deliver multiple output objects (such as an HTML page with associated GIF or PNG images) to a client application. Transient package output is available on both stored process and workspace servers, but the implementations differ. On the stored process server, transient package output is indicated when _RESULT is set to PACKAGE_TO_ARCHIVE and the input parameter _ARCHIVE_PATH is set to TEMPFILE . On the workspace server, transient package output is indicated when _RESULT is set to PACKAGE_TO_REQUESTER .
- Permanent package output creates a package in a permanent location on a WebDAV server or in the server file system. The package is immediately accessible to the stored process client, but is also permanently accessible to any client with access to WebDAV or the server file system. This result type is a convenient way to publish output for permanent access. Output to WebDAV is indicated when _RESULT is set to PACKAGE_TO_WEBDAV . The input parameter _COLLECTION_URL contains the target location. The input parameters _HTTP_USER and _HTTP_PASSWORD might be set if the WebDAV server is secured and credentials are available. The _HTTP_PROXY_URL parameter is set if an HTTP proxy server is required to access the WebDAV server. Output to the server file system is indicated when _RESULT is set to PACKAGE_TO_ARCHIVE . The input parameters _ARCHIVE_PATH and _ARCHIVE_NAME contain the target repository and filename, respectively.
  
  Permanent package output can also be published to a channel or to an e-mail recipient. For more information about the parameters that are used for publishing packages, see Advanced Package Publishing.

Note: Although the result type is chosen when you define a stored process, the result type can be changed by the client application through calls to the Stored Process Service API. Where possible, it is recommended that you write stored processes to support any appropriate client result type. This enables a client application to select the result type most appropriate for that application. The program can determine the desired client result type by examining the _RESULT input parameter. The %STPBEGIN and %STPEND macros include support for any of the four result types. For more information, see Using the %STPBEGIN and %STPEND Macros. The following stored process is capable of generating streaming, transient package, or permanent package output. (It can also be run with _RESULT set to STATUS , but this would produce no useful result.)

 *ProcessBody;
 %stpbegin;
 proc print data=SASHELP.CLASS noobs;
 var name age height;
 run;
 %stpend;

[cautionend]

The input parameters that were mentioned previously are set by the stored process client APIs and are reserved parameters. They cannot be overridden by passing in new values through the normal parameter interface. Special API methods are provided to set the result type and associated parameters for a stored process. For more information about specific input parameters, see Using Reserved Macro Variables. For more information about developing stored processes that product package results, see Developing Stored Processes with Package Results.

Top of Page