Chapter Contents

Previous

Next

Creating C++ Programs under CMS

Before running the translator, compiler, COOL, or any C++ program, ensure that the transient library is available on an accessed minidisk or that it is installed in a segment available to your virtual machine. Consult your SAS Installation Representative to determine if this has been done for you.

Translating and Compiling Your Program under CMS

The LCXX EXEC invokes the translator and compiler. Optionally, you can also invoke the OMD370 object module disassembler utility. The LCXX EXEC has the following format:

LCXX fileid <(options<)>>

where fileid is the fileid of the file to be translated and compiled. The fileid can name a file on a CMS minidisk, a file in the XEDIT ring, or a file in a Shared File System (SFS) directory. The format of the fileid is described in Specifying a fileid. The options are any translator options, compiler options, or, if the OMD option is used to invoke the OMD370 utility, OMD370 options. See Option Summary for the options that are available for use with the translator.

By default, the translator writes its output (C code) to a temporary file and runs the compiler if no errors were found. The name of the output file has the same filename as the input file and a filetype of TROUT. If the input file is on an accessed minidisk or directory accessed as a minidisk, the temporary output file is written on that minidisk. If the minidisk is not write-accessed, the output file is written to the minidisk accessed as filemode A. If the input file is in an SFS directory that is not accessed as a minidisk, the output file is written to that directory. If the directory is not writable, the output file is written to your top directory. If the input file is in the XEDIT ring, the output file is written to the minidisk accessed as filemode A.

Note:    If you have an existing file with the same name as the C output file, it is overwritten and then erased. Therefore, you may want to use the savec option to specify a different name for the output file, as discussed in the following section.  

Saving the output file under CMS

You can use three options to save the intermediate code produced by the translator: pponly , tronly , and savec . Here is a brief description of each of these options:

pponly	performs only the preprocessing step. The program is not translated or compiled. The output file contains C++ code.
tronly	performs the preprocessing and translation steps. The program is not compiled. The output file contains C code.
savec	saves the C code resulting from translation. The program is preprocessed, translated, and compiled. The output file contains C code. Under CMS, the savec option must be the last option on the command line. LCXX uses the remainder of the command line as the fileid of the output file.

For detailed information on these three options, refer to Option Descriptions.

Note:    Because both tronly and savec save the intermediate C code file, you should use only one of them. For example, if you use tronly , you do not need to also specify savec , unless you want to rename the intermediate C code file.  

Specifying a fileid

The following sections illustrate specifying fileids (for both input and output files). The examples are divided according to your file environment, such as files on accessed minidisk or directory, files on the XEDIT ring, and files in a Shared File System (SFS) directory.

Files on an accessed minidisk or directory

To specify a file on an accessed minidisk or directory, use

filename <filetype <filemode>>

Optionally, you can separate the filename, filetype, and filemode by periods instead of blanks. If the file is on an accessed minidisk or directory, you can optionally prefix the fileid with cms: . If the filetype is omitted, the default filetype for the input file is CXX. If the filemode is omitted, the translator searches all accessed minidisks and directories. Here are some examples:

LCXX MYPROG

The input file is MYPROG CXX *. The output file is temporary and is named MYPROG TROUT. The output file is written to either the minidisk where the input file was found or to the A disk if the input minidisk is not writable.

LCXX MYPROG.CPP

The input file is MYPROG CPP *. The output file is temporary and is named MYPROG TROUT.

LCXX MYPROG (SAVEC MYPROG CPPOUT B2

The input file is MYPROG CXX *. The savec option stores the output file as MYPROG CPPOUT B2.

LCXX CMS:MYPROG.CPP.J (NOWARN SAVEC CMS:MYPROG.CPPOUT.B2

The input file is MYPROG CPP J. The nowarn option suppresses translator warning messages. The savec option stores the output file as MYPROG CPPOUT B2. Note that the savec option must be the last option specified.

Files in the XEDIT ring

If you run the LCXX EXEC from the XEDIT command line and the input file is in the XEDIT ring, the translator automatically reads the input file from XEDIT. However, the translator does not write its output to an XEDIT file.

Files in an SFS directory

To specify a file in an SFS directory, use the following format:

sf: filename <filetype <directory-name | namedef>>

The sf: prefix is required. If the filetype is omitted, the default filetype is CXX. If you do not specify either a directory-name or a NAMEDEF, the default is your top directory. Here are some examples:

LCXX SF:MYPROG

MYPROG CXX

is the input file.

MYPROG TROUT

is the temporary output file.

LCXX SF:MYPROG CPP

MYPROG CPP.

MYPROG TROUT

is the temporary output file.

LCXX SF:MYPROG (SAVEC MYPROG CPPOUT

MYPROG CXX.

is the input file.

MYPROG CPPOUT.

is the permanent output file, created by the savec option. Note that the output file prefix defaults to the input file prefix.

LCXX SF:MYPROG CXX CXXPROG

MYPROG CXX CXXPROG

is the input file, where CXXPROG is a NAMEDEF defined in a NAMEDEF command.

MYPROG TROUT.

is the temporary output file.

LCXX SF:MYPROG CXX .CXX.PROJ (SAVEC CMS:MYPROG TROUT B2

MYPROG CXX .CXX.PROJ

is the input file.

MYPROG TROUT B2

is the permanent output file, created by the savec option. The cms: prefix is used to override the default prefix of sf: .

Locating header files under CMS

The translator searches for header files in one of two locations, depending on how the filename is specified in the #include statement. When the filename is surrounded by angle brackets in the #include statement, the translator assumes that the header file is a member of a macro library that has been designated GLOBAL. The syntax for this form of the #include statement is

#include <member.ext>

The .ext part can be omitted.

For example:

#include <new.h>

In this case, the translator includes the first member named NEW that it finds in the GLOBAL macro libraries.

When the filename is surrounded by double quotes in the #include statement, the translator searches for the specified file. The syntax for this form of the #include statement is

#include "fileid"

fileid can be any CMS fileid in the cms: , xed: , or sf: formats. For example, suppose you had the following #include statement in your program:

#include "project.h"

In this case, the translator attempts to include the file named PROJECT H *. If this file cannot be found, the translator searches the macro libraries that have been designated GLOBAL. The translator uses that part of the fileid preceding any blank or period as the member name. Refer to the SAS/C Library Reference, Third Edition, Volume 1 for more information about CMS filename formats.

If you are using the translator in a version of CMS that supports the Shared File System, the translator automatically searches your top directory for header files before searching any accessed minidisks. You can specify additional directories to be searched by using the _HEADERS environment variable. Refer to the SAS/C Compiler and Library User's Guide for more information about using environment variables.

Before invoking the LCXX EXEC, be sure that the C++ standard header files and C standard header files are available. The C++ standard header files are in a macro library named LCXX370 MACLIB. The C standard header files are in a macro library named LC370 MACLIB. To make these header files available to the translator, issue the following command before calling the translator:

GLOBAL MACLIB LCXX370 LC370

If your program requires header files in another macro library, add the filename of the macro library to the GLOBAL MACLIB command.

CXXMACLIBS and CXXOPTIONS GLOBALV variables

The LCXX EXEC automatically issues a GLOBAL MACLIB command for you if you set the CXXMACLIBS environment variable to the names of the macro libraries you want to be made available. In addition, if you set the CXXOPTIONS environment variable to a list of translator options, LCXX automatically passes those options to the translator in addition to the options you specify on the command line. Set these variables using the GLOBALV command. Both variables should be in the LC370 group.

The GLOBALV command (see CXXMACLIBS GLOBALV Command) sets the CXXMACLIBS variable so that the LCXX EXEC automatically issues a GLOBAL MACLIB command for the C++ standard header files and the C standard header files.

CXXMACLIBS GLOBALV Command


GLOBALV SELECT LC370 SETLP CXXMACLIBS LCXX370 LC370

You can suppress the use of environment variables for a single compilation by specifying the NOGLOBAL option on the LCXX command line.

The following example shows how to set the CXXOPTIONS variable to cause LCXX to automatically pass the overload option to the translator:

GLOBALV SELECT LC370 SETP CXXOPTIONS OVERLOAD

Options specified on the command line override options specified via the CXXOPTIONS variable.

Linking Your Program under CMS

All C++ programs must be preprocessed by the COOL object code preprocessor because COOL automatically creates object code used during the initialization and termination of static objects. In addition, COOL supports linking object files that contain mixed case external names longer than eight characters, such as those created by the compiler for source code generated by the translator. Such external names cannot be handled by the linkage editor unless the object code has been preprocessed by COOL.

COOL EXEC

The COOL EXEC calls COOL to preprocess your object code, and it optionally invokes the CMS GENMOD command. The format of the COOL EXEC is

COOL <filename1 <filename2 . . .>> (CXX <options<)>>

Filename1, filename2, and so on are the filenames of primary input files or AR370 archives. For each filename in the list, COOL first checks for an AR370 archive with this filename; if one is not found, it looks for a TEXT file. For instance, if the filename FINANCE is specified, COOL first looks for FINANCE A, an AR370 archive on an accessed minidisk or in an SFS directory accessed as a minidisk. If FINANCE A is not found, COOL looks for FINANCE TEXT. Any TEXT files specified on the command line may contain object code, COOL control statements, or both. Any AR370 archives specified on the command line are used to resolve unresolved references during the processing of the other input files.

For example, the following command line causes COOL to use MYPROG TEXT as its only input file, provided that MYPROG A is not found:

COOL MYPROG (CXX

If you do not specify any filenames, COOL prompts you for filenames. Enter as many filenames as necessary in response to the COOL: prompt. Enter a null string (that is, press the ENTER key) to cause COOL to begin processing the input files.

The CXX option is the only required option. This option causes the standard C++ object library to be added to COOL's autocall list. Refer to COOL Options for more information about COOL options. Before calling COOL, you must issue a GLOBAL TXTLIB command to make the standard C object libraries available for autocall resolution. The standard C object libraries are LC370BAS TXTLIB and LC370STD TXTLIB.(footnote 1) The following GLOBAL TXTLIB command makes these libraries available for COOL:

GLOBAL TXTLIB LC370BAS LC370STD

The COOL EXEC automatically issues a GLOBAL TXTLIB command for you if you specify the names of the object libraries in the TXTLIBS environment variable.

The COOL input files can contain control statements instead of, or in addition to, object code. For more information about COOL control statements, refer to COOL Control Statements.

COOL writes the preprocessed object code to a file named COOL370 TEXT A1. This file can be used as input to the LOAD command or the LKED command.

The COOL listing file is written to the terminal by default. You can redirect it to a disk file. Enter the redirection argument in the filename part of the command line, not the options part. For example, the following command causes COOL to write its listing file to MYPROG COOLMAP A1:

COOL MYPROG >MYPROG.COOLMAP.A1 (CXX

There are many special considerations for linking C++ programs under CMS that parallel the considerations for linking C programs. Refer to the SAS/C Compiler and Library User's Guide for a complete discussion of these considerations.

Creating a MODULE file

The COOL EXEC accepts the GENMOD option, which specifies that the EXEC should load the COOL370 TEXT file and issue the GENMOD command. You can specify the name of the MODULE file following the GENMOD option. If you do not specify a name, the COOL EXEC uses the first input filename on the command line as the name of the MODULE. For example, the following command creates MYPROG MODULE:

COOL MYPROG LIBFNC (CXX GENMOD

The following command creates APPL1 MODULE:

COOL MAIN1 SUB1 SUB2 (CXX GENMOD APPL1

You can also cause COOL to invoke the START command or the LKED command. Refer to the SAS/C Compiler and Library User's Guide for more information.

Running Your Program under CMS

C++ programs can be run just like any other program under CMS. The most frequently used method for running a program is to create a MODULE file (as shown previously) and then call the module as a CMS command. For example, the following command calls MYPROG MODULE, passing the program option -z , the run-time option =48k , and redirecting stdin to INPUT FILE *:

MYPROG -z =48k <INPUT.FILE

You can also load a TEXT file and use the START command to run it. For example, suppose you have created MYFILE TEXT by using COOL to preprocess MYPROG TEXT. The following commands can be used to load and run it, passing the same options as in the previous example:

LOAD MYFILE
START * -z =48k <INPUT.FILE

Programs that can be called from the CMS EXEC processor should be prepared to accept tokenized parameters. This form of parameter is at most eight characters long and is translated to uppercase by CMS. Refer to the SAS/C Compiler and Library User's Guide for more information. If you want to run your program with the SAS/C Debugger, use the '=D' option when you call your program, as in the following example:

MYPROG =D

For more information on debugging C++ programs, refer to Debugging C++ Programs Using the SAS/C Debugger.

Chapter Contents

Previous

Next

Top of Page

Copyright © 2001 by SAS Institute Inc., Cary, NC, USA. All rights reserved.