GBOOCH@USC-ISIF.ARPA (Grady Booch) (03/28/86)
Some time ago, there was net traffic that weighed the merits of judging the size of an ada program by counting actual lines of code versus the number of semicolons. I recently gathered these statistics on some software I've written, and came up with the following statistics: number of units: 892 loc: : 123584 semis : 60547 this works out to be 2.04 lines of code per semicolon. Has anyone else analyzed a substantial amount of software? I'm interested in seeing if this ratio holds across different application domains, or if it is just a reflection of my personal coding style. egb -------
RCONN@SIMTEL20.ARPA (Rick Conn) (03/29/86)
Hello, Grady, Re your question on the ratio of LOC/semicolon, I pulled some files from various contractors and projects from the repository and ran them thru File Checker. While this is just a small sample, it is interesting. Perhaps when I get around to porting/rewriting File Checker to SIMTEL20, we can run it against the entire repository. FILE CHECKER, Version 1.3 File Name Statements Comments Stmts+Cmts Lines ============================== ========== ========== ========== ========== abstract.src 4427 6015 10442 17975 spell2.src 2743 2710 5453 53334 tcpstand.src 217 88 305 473 constrct.src 530 412 942 4148 createtb.src 661 293 954 1546 a970.src 654 592 1246 3516 [contr13.sources]fcheck.src 153 252 405 594 [contr13.sources]help.src 539 444 983 2260 [contr13.sources]list.ada 161 211 372 498 [contr13.sources]pager.src 741 533 1274 3236 ** TOTALS ** 10826 11550 22376 87580 Rick -------
RCONN@SIMTEL20.ARPA (Rick Conn) (03/29/86)
By the way, in my last message, "Statements" is a semicolon count (intelligent since it knows about semis in parameter lists, etc, and does not count them) and "Lines" is a new line count. "Comments" are comment lines (first non- blank is --). Rick -------