Axis Computer Consulting Services, Inc.

I have uploaded a new version of my SQR source code analyzer perl script [here]. This version adds support for the #IF statements (eg. #IF {TEST} = 'Y'). Also added is a new command line option (-sk) that allows you to specify procedure names NOT TO trace when doing the analysis of the source code (one SQR I analyzed called a procedure 24,599 times).
The program that I was having problems analyzing got me thinking about performance, and how I could improve it, so after adding the "skip" code to the program, I profiled this "nasty" SQR. In total, this SQR includes 67 include files and runs over 40,000 lines of code. I ran the Perl profiler on the script as it analyzies this SQR (perl -d:DProf sqr-analyzer.pl...) and the following is the output of the profiler information (dprofpp tmon.out):

Total Elapsed Time = 9652.995 Seconds User+System Time = 1349.652 Seconds Exclusive Times
%Time	ExclSec	CumulS	#Calls	sec/call	Csec/c	Name
98.3	1326.	19012.	46976	0.0282	0.4047	main::process_function_call
0.76	10.27	9.709	190126	0.0001	0.0001	main::print_function_tree_indent
0.63	8.535	19028.	71575	0.0001	0.2659	main::process_function_call_line
0.27	3.608	2.892	238931	0.0000	0.0000	main::clean_line
0.26	3.451	4.265	62828	0.0001	0.0001	main::process_preproc_if
0.16	2.114	1.900	71575	0.0000	0.0000	main::remove_from_call_tree
0.10	1.332	1.117	71575	0.0000	0.0000	main::add_to_call_tree
0.07	0.995	1.663	68	0.0146	0.0245	main::source_scan
0.03	0.400	0.400	68	0.0059	0.0059	main::get_file_location
0.01	0.170	0.170	1	0.1700	0.1700	main::print_function_list_by_sqc
0.00	0.060	0.060	1	0.0600	0.0600	main::print_define_list_by_SQC
0.00	0.050	0.050	670	0.0001	0.0001	main::process_procedure
0.00	0.010	0.010	1	0.0100	0.0100	main::print_function_list_by_function
0.00	0.010	0.008	670	0.0000	0.0000	main::add_to_function_list
0.00	0.010	0.010	1	0.0100	0.0100	main::print_function_call_stats

The process using the most time is running at a little over .02 seconds per call, I am not sure I can improve this too much. This program that I profiled took the longest by far (2 hrs, 40 min on a P4 1.8GHz machine under Cygwin, 50 minutes on a Sun Ultra server with 12 CPU's).