When writing your program, you can conveniently test statements as you go along using the REXXTRY command from the OS/2 command prompt. REXXTRY is a kind of REXX mini-interpreter that checks REXX statements one at a time. If you run it with no parameter, or with a question mark as a parameter, REXXTRY will also briefly describe itself.
From your current OS/2 window, just open another window and from the command prompt type:
rexxtry
REXXTRY will describe itself and ask you for a REXX statement to test. Enter your statement; REXXTRY will run it and return any information, or display an error message if a problem is encountered. REXXTRY remembers any previous statements you have entered during the session, so to continue just type the next line in your program and REXXTRY will check it for you.
Enter an equal sign (=) to repeat your previous statement, or a question mark (?) to invoke system-provided online information about the REXX language.
When you're done, type:
exit
and press Enter to leave REXXTRY.
You can also enter a REXX statement directly on the command line for immediate processing and exit:
rexxtry call show
In this case, entering CALL SHOW displays the user variables provided by REXXTRY.
Whereas REXXTRY runs and debugs one REXX statement at a time, you can use PMREXX to do the same for an entire REXX procedure. PMREXX is a windowed application that shows, in a PMREXX window, any results displayed by your REXX procedure. You can scroll through the window using its scroll bar to view all the output.
PMREXX also provides a single line input field, so that you can supply input to the REXX procedure or to any commands called by it.
PMREXX is an OS/2 installation option; it is available if you select it during OS/2 installation. Assuming you've done this and PMREXX is available, you start it by entering PMREXX from an OS/2 command prompt, followed by the name of a REXX procedure you want to run; for example:
pmrexx myprog
You can also supply arguments for the REXX procedure. In the next example, PMREXX will run the procedure MYPROG and pass MYPROG the argument c:\test\test.data:
pmrexx myprog c:\test\test.data
PMREXX will run the REXX procedure and display any output generated by it in a scrollable output box that is part of the PMREXX window. If the procedure prompts for input, the prompt is displayed in the output box. You would then type the response in the smaller input box near the top of the PMREXX window.
To try it, start an OS/2 session and create this REXX procedure:
/* DIRTEST.CMD -- displays directory contents */ do forever say 'Enter the name of a directory' parse upper pull response if response='QUIT' then leave 'dir' response end exit
Then start PMREXX again for the DIRTEST procedure:
pmrexx dirtest
When the PMREXX window is displayed, you'll see the prompt in the output box. Position the cursor in the input box, type the name of a directory, and press Enter. The contents of the directory will be listed in the output box, and another prompt will be displayed.
When the amount of displayed output exceeds the size of the output box, a slider box appears in the scroll bar.
To end the DIRTEST program, type:
quit
in the input box and press Enter.
When you run your REXX programs with PMREXX, you can use a function called RxMessageBox to display messages. This example uses it to display an error message:
/* ERROR.CMD -- Check the input parameter */
arg count
if count~datatype('Whole') <> 0 then do
RxMessageBox("Argument" count "is not a whole number")
exit
end
RxMessageBox displays a Presentation Manager message box titled Error! and an OK button. The REXX program waits until you click OK.
You can change the message box title and buttons, and add a colorful icon to the message box:
/* FILECHK.CMD -- Does the file exist? */
if file~Query('Exists') <> ''
then do
reply = RxMessageBox("Do you want to replace file",,
"Replace File?", "YesNo", "Question")
if reply = 7 then exit /* user pressed 'No' */
end
FILECHK.CMD displays a question in a message box with a question-mark icon and two buttons, labeled Yes and No. Clicking Yes or No gives your program the number of the button you chose. (Yes is 6 and No is 7.)
To use PMREXX as a miniature development environment for your procedures, run PMREXX in one OS/2 session and an editor in another session. Use the editor to change and save the procedure being developed. Then switch to the PMREXX session to run the modified procedure.
You can restart a procedure from PMREXX by selecting Trace from the action bar in the PMREXX window. If the procedure is running, select Halt procedure to stop it, then select Restart from the menu. Otherwise, just select Restart. The latest version of the procedure will run again.
PMREXX includes several functions useful for debugging your procedures. You can, for instance, start an interactive trace from PMREXX without having to add a TRACE instruction to your procedure. Select Trace on the action bar, and then select Interactive trace on. A check mark on the menu shows that interactive tracing is on. To stop the interactive trace, just select Interactive trace on again. One advantage of using the interactive trace from PMREXX is that you can turn the trace on and off while the program is running.
Once tracing is active, you can step through your procedure one clause at a time, re-do a clause that was just processed, or enter a line of REXX clauses for immediate processing. The ability to enter REXX clauses is especially useful if you want to try a fix to a problem interactively or if you want to test instruction paths that are otherwise difficult to trigger.
For example, you might want to test some error handling instructions, but cannot easily create the condition that would cause the error. By using the interactive trace, you can add REXX instructions at the right moment to fake the conditions that would cause the error handling instructions to be processed.
To process a line of one or more REXX clauses, type the line in the PMREXX input box when tracing is active and press Enter. For example, you could enter:
do i=1 to 10; say 'hello' i; end
The line is processed before the next REXX clause in the program is processed. If you simply want to step ahead to the next clause, press Enter without typing anything in the input area. You can also step ahead by selecting Trace next clause from the Trace menu. If you want to process the last REXX clause again, select Re-do last clause from the Trace menu.
To stop tracing, select Trace off from the Trace menu. This item is not selectable when the REXX procedure is waiting for user input. In this case select Interactive trace on again to stop the trace.