Using Unnamed and Named Pipes under Windows |
Introduction to Named Pipes |
The named pipes capability is one of the most powerful tools available in SAS under Windows for communicating with other applications. The named pipes feature enables bidirectional data or message exchange between applications on the same machine or applications on separate machines across a network. The following figure illustrates these two methods of communication.
Communication Using Named Pipes
The applications can be SAS sessions or other Windows applications. For example, you can use the PRINTTO procedure to direct the results from SAS procedures to another Windows application, using a named pipe. Therefore, you have the choice of having multiple SAS sessions that communicate with each other or one SAS session communicating with another Windows application.
Whether you are communicating between multiple SAS sessions or between a SAS session and another Windows application that supports named pipes, the pipes are defined in a client/server relationship. One process is defined as the server, while one or more other processes are defined as clients. In this configuration, you can have multiple clients send data to the server or the server send data to the various clients. Named pipes enable you to coordinate processing between the server and clients using various options.
Named Pipe Syntax |
You can use a named pipe anywhere you use a fileref in SAS. To use a named pipe, issue a FILENAME statement with the following syntax:
FILENAME fileref NAMEPIPE 'pipe-specification' <named-pipe-options>; |
You can use the following arguments with this syntax of the FILENAME statement:
is any valid fileref as described in Referencing External Files.
is the device-type keyword that tells SAS you want to use a named pipe.
This argument has two mutually exclusive syntaxes:
indicates you are establishing a pipe on a single PC or defining a server pipe across a network. The pipe-name argument specifies the name of the pipe.
indicates you are establishing a client pipe over a network named-pipe server. Remember to include the double backslash (\\) in this situation. The pipe-name argument specifies the name of the client pipe. The server-name argument specifies the name of the named-pipe server.
can be any of the following. The default value is listed first:
indicates whether the client or server is to wait for data to be read if no data are currently available. BLOCK indicates to wait and is the default. NOBLOCK indicates not to wait. Control is returned immediately to the program if no data are available in the pipe. Writing to the pipe always implies BLOCK.
indicates the type of pipe. BYTE is the default. The difference between a BYTE pipe and a MESSAGE pipe is that a MESSAGE pipe includes an encoded record length, whereas a BYTE pipe does not.
indicates the amount of time the client or server is to wait to establish the pipe. The minimum value for seconds is 10. This option allows time for synchronization of the client and server. The default waiting period is 10 seconds.
There are two values for the seconds argument that indicate special cases:
is valid only when defining the server and indicates that if an end-of-file (EOF) is received from a client, the server is to try to connect to the next client.
All of these options are consistent with terminology used in Windows programmers' reference guides such as those options provided with the Microsoft Win32 SDK.
Using the CALL RECONNECT Routine |
A special SAS CALL routine, CALL RECONNECT, enables the server to disconnect the current client and try to connect to the next available client. Normally, a pipe terminates when the client side of the pipe sends an end-of-file to the server. To break the pipe connection at any time, the server SAS session can issue a CALL RECONNECT statement. For an illustration of this routine, see The CALL RECONNECT Routine.
Using Named Pipes in SCL |
To establish named pipes using SCL code, you must use the FOPEN function to open a file (or pipe) before you can access it. In doing so, you must specify the appropriate open mode for both the client and server applications so that the two can communicate over the pipe. Here is a summary of the different nodes you can use:
If the server accesses the pipe as... | then the client must access it as... |
---|---|
I (input) | O (output) |
O (output) | S (sequential) |
U (update) | O (output) or S (sequential) |
Named Pipe Examples |
The best way to understand named pipes is to examine several different examples illustrating their use. In most of the examples in this section, the named pipe is established between two SAS sessions. However, named pipes work between SAS and other applications that support named pipes.
The simplest named pipe configuration is one server connected to one client, as shown in One Server Connected to One Client.
One Server Connected to One Client
In the following example, a named pipe called WOMEN is established between two SAS sessions. The server SAS session selectively sends data to the client SAS session. You can start the server or the client first; one waits 30 seconds for the other to connect.
In the first SAS session, create a named pipe as a server:
/* Creates a pipe called WOMEN, acting */ /* as a server. The server waits 30 */ /* seconds for a client to connect. */ filename women namepipe '\\.\pipe\women' server retry=30; /* This code writes three records into */ /* the named pipe called WOMEN. */ data class; input name $ sex $ age; file women; if upcase(sex)='F' then put name age; datalines; MOORE M 15 JOHNSON F 16 DALY F 14 ROBERTS M 14 PARKER F 13 ;
In the second SAS session, you can use SAS statements to exchange data between the two SAS sessions. For example, you can submit the following program from the client session:
/* Creates a pipe called WOMEN, acting */ /* as a client. The client waits 30 */ /* seconds for a server to connect. */ filename in namepipe '\\.\pipe\women' client retry=30; data female; infile in; input name $ age; proc print; run;
The following program is another example of a single client and server. This example illustrates using the PRINTTO procedure to direct results from the SUMMARY procedure to another Windows application, using a named pipe called RESULTS:
filename results namepipe '\\.\pipe\results' server retry=60; proc printto print=results new; run; proc summary data=monthly; run;
You can choose to have one server connected to several clients. In this case, the named pipe configuration looks like the configuration shown in One Server Connected to Several Clients.
One Server Connected to Several Clients
In this configuration, the data connection is initially between the server and the first client. When this connection is terminated, the server connects to the second client, and so on. The connection can return to the first client after the last client's connection is broken if your program is set up to do so.
You must use the EOFCONNECT option to cause the connection to move properly from one client to the next. Here is an example of using the EOFCONNECT option with one server SAS session and two clients. The clients can be on the same PC or on a PC connected across a network.
In the first SAS session, submit the following statements:
/* Creates a pipe called SALES, acting */ /* as a server. The server waits 30 */ /* seconds for a client to connect. */ /* After the client has disconnected, */ /* this server SAS session tries to */ /* connect to the next available client */ filename daily namepipe '\\.\pipe\sales' server eofconnect retry=30; /* This program reads in the daily */ /* sales figures sent from each client.*/ data totsales; infile daily; input dept $ item $ total; run;
In the second SAS session, submit the following statements:
/* Creates a pipe called SALES, acting */ /* as a client. The client waits forever */ /* for a server to connect. After the */ /* first client has disconnected, the */ /* second client connects with the server.*/ /* The first client is the TOYS dept. */ filename dept1 namepipe '\\.\pipe\sales' client retry=-1; data toys; input item $ total; dept='TOYS'; file dept1; put dept item total; datalines; DOLLS 100 MARBLES 10 BLOCKS 50 GAMES 60 CARS 40 ; /* The second client is the SPORTS dept.*/ /* These data could come from a separate */ /* SAS session. */ filename dept2 namepipe '\\.\pipe\sales' client retry=-1; data sports; input item $ total; dept='SPORTS'; file dept2; put dept item total; datalines; BALLS 30 BATS 65 GLOVES 15 RACKETS 75 FISHING 20 TENTS 115 HELMETS 45 ;
In the following example, the NOBLOCK option is used to specify that if no data are available when the pipe is read, then the program should continue performing. If the default value of BLOCK had been used, then the pipe would wait indefinitely until data were found in the pipe. The EOFCONNECT option is used to tell the server that when a client sends an end-of-file, the server can connect with a new client. The RETRY= option tells the server to look for any new clients for 20 seconds while the client waits indefinitely on a server. The clients can be on the same PC or on a PC connected across a network. A server connects to one client at a time, and the clients queue in a serial order waiting to connect to the server.
In the first SAS session, submit the following statements:
/* Defines a named pipe called LINE. */ /* Use the NOBLOCK option to specify */ /* that if no data are available when */ /* the read is performed, then continue.*/ /* Use the EOFCONNECT option to tell */ /* the server to try to connect with a */ /* new client if an end-of-file is */ /* encountered. Use the RETRY= option */ /* to tell the server to look for any */ /* new clients for 20 seconds. */ filename data namepipe '\\.\pipe\line' server noblock eofconnect retry=20; /* This DATA step reads in all data */ /* from any clients connected to the */ /* named pipe called LINE. */ data all; infile data length=len; input @; /* If the length of the incoming */ /* record is 0, then no data were */ /* found in the pipe; otherwise, */ /* read the incoming data. */ if len ne 0 then do; input machine $ width weight; output; end; run; proc print; run;
Each of the following DATA steps below can be carried out on several PCs connected across a network:
/* Defines a named pipe called LINE. */ /* The RETRY= option is set such that */ /* the clients wait forever until a */ /* server is available */ /* (that is, RETRY=-1). */ filename data namepipe '\\.\pipe\line' client retry=-1; /* This is information from the */ /* first machine/client. */ data machine1; file data; input width weight; machine='LINE_1'; put machine width weight; datalines; 5.3 18.2 3.2 14.3 4.8 16.9 6.4 20.8 4.3 15.4 6.1 19.5 5.6 18.9 ; /* This is information from the */ /* second machine/client. */ filename data namepipe '\\.\pipe\line' client retry=-1; data machine2; file data; input width weight; machine='LINE_2'; put machine width weight; datalines; 4.3 17.2 5.2 18.4 6.8 19.9 3.4 14.5 5.3 18.6 4.1 17.1 6.6 19.5 ;
The following example demonstrates how to set up a named pipe server to establish a connection with two clients. (For this example, you need three active SAS sessions.) In this example, the CALL RECONNECT routine is used to reconnect to the next client on the named pipe if it has been at least 30 seconds since the previous client has sent any data. Each client is a data entry operator, sending data to the server SAS session.
In the server SAS session, submit the following statements:
filename data namepipe '\\.\pipe\orders' server noblock eofconnect retry=30; data all; infile data length=len missover; input @; /* If the length of the incoming */ /* record is 0, then no data were */ /* found in the pipe; otherwise, */ /* read the incoming data */ if len ne 0 then do; input operator $ item $ quantity $; if item='' or quantity='' then delete; else output; put operator= item= quantity=; end; /* If no data are being transmitted,*/ /* try reconnecting to the next */ /* available client. */ else do; /* Use the named pipe fileref */ /* as the argument of */ /* CALL RECONNECT. */ call reconnect('data'); end; run;
In the second SAS session, which is the first data entry operator, submit the following statements:
filename data namepipe '\\.\pipe\orders' client retry=-1; data entry1; if _n_=1 then do; window entry_1 #1 @2 'ENTER STOP WHEN YOU ARE FINISHED' #3 @5 'ITEM NUMBER - ' item $3. #5 @5 'QUANTITY - ' quantity $3.; end; do while (upcase(_cmd_) ne 'STOP'); display entry_1; file data; put 'ENTRY_1' +1 item quantity; item=''; quantity=''; end; stop; run;
In the third SAS session, which is the second data entry operator, submit the following statements:
filename data namepipe '\\.\pipe\orders' client retry=-1; data entry2; if _n_=1 then do; window entry_2 #1 @2 'ENTER STOP WHEN YOU ARE FINISHED' #3 @5 'ITEM NUMBER - ' item $3. #5 @5 'QUANTITY - ' quantity $3.; end; do while (upcase(_cmd_) ne 'STOP'); display entry_2; file data; put 'ENTRY_2' +1 item quantity; item=''; quantity=''; end; stop; run;
Copyright © 2010 by SAS Institute Inc., Cary, NC, USA. All rights reserved.