On most Unix systems, the programs prof(1), gprof(1), and analyze_coverage(1) can be used to profile your Modula-3 program's performance or to analyze which source lines of your program have been executed. However, in order to use these tools, you must compile your program with certain compiler options.
To profile your program for use with prof(1) or gprof(1), add the following compiler option to your m3makefile:
option("profiling","T")
This will cause your program to be compiled and linked for profiling. If you have already built your program without profiling, and you want to turn profiling on, delete the object files and executable program and rebuild.
When you run your program, a file named "mon.out" (for prof(1)) or "gmon.out" (for gprof(1)) will be created. To view the profile, run the prof(1) or gprof(1) program using the command-line:
prof prog mon.out gprof prog gmon.out
where prog is the filename of your program's binary.
To determine how many times each source line and procedure of your Modula-3 program were exececuted on a particular run, you need to compile it to be used with the analyze_coverage(1) tool. To do this, add the following compiler option to your m3makefile:
option("coverage","T")
Coverage data will be available only for those modules that have been compiled with this option, so if you want coverage analysis for all your sources and you have just turned this option on, you will need to delete your derived directory and recompile everything. To produce the coverage data, you must also link your program with this option.
Once you've compiled your program for coverage analysis, just run it. By default, this will produce a file named "coverage.out" in the current directory (you can set the environment variable COVERAGE_DATABASE to the name of a different file if you prefer). This file is the coverage database. The coverage information is accumulated in this file from run to run, so if you want to see what your program does on a single run, you should delete the coverage database file before running your program. For the same reason, the profiling data file is rather large; furthermore, as it is augmented by each execution of the program, you may want to compress it from time to time (see analyze_coverage(1) for more details).
Once you've run your program, you can use the analyze_coverage(1) tool to show you what's in the coverage database. For this tool to work, you must point it to the directories containing your source files. You can do this by setting the COVERAGE_PATH environment variable, or by supplying the "-S" option to analyze_coverage(1). Again, see the analyze_coverage(1) man page for details.
Note that because of the extensive data collection performed by this mode of profiling, the execution time of the program can be significantly larger when it is enabled; thus, simultaneous time profiling can produce erroneous results.