First, things that you do to trick a compiler into generating more efficient code can result in less efficient code. Second, comments may be essential but so are good names and maintainable code. If you want to have the people who update your code to bless you, use names instead of magic numbers. Approximations to constants lie Pi and e should have suitable names in case someone needs more precision. more precision.
-- Shmuel (Seymour J.) Metz http://mason.gmu.edu/~smetz3 ________________________________________ From: IBM Mainframe Discussion List <IBM-MAIN@LISTSERV.UA.EDU> on behalf of Bob Bridges <robhbrid...@gmail.com> Sent: Sunday, September 6, 2020 6:29 PM To: IBM-MAIN@LISTSERV.UA.EDU Subject: Re: Constant Identifiers To tell you the truth, I haven't written for a compiler in long enough that I don't have any idea which ones do what. I wouldn't trust REXX, CLIST or VBA to do it; I try to assume the worst, and make such things explicit. Actually I might not trust a compiler to do it, either, even if I knew that you're right. I'm just not very trusting about such things. (And it's not ~very~ obfuscatory. In real life I'd include a comment, something like this: SphereVolConst=4/3*3.14159E0 /* for calculating the volume of a sphere inside the loop */ That looks like a lot of documentation, to some programmers. But the older I get, the more comments I write into even fairly simple programs.) --- Bob Bridges, robhbrid...@gmail.com, cell 336 382-7313 /* I'm told that Wagner's music is not as bad as it sounds. -Mark Twain */ -----Original Message----- From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On Behalf Of Robin Vowels Sent: Saturday, September 5, 2020 23:09 That's obfuscation. If it's in a loop, I'd expect the compiler to move the constant part outside of the loop, or to evaluate the constant part at compilation time. In any case, the more efficient form eliminates the division 4/3, avoiding the avaluation in fixed-point form (and the conversion to floating-point). -----Original Message----- From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On Behalf Of Seymour J Metz Sent: Saturday, September 5, 2020 21:47 Don't most compilers these days do constant folding? ----- Original Message ----- From: "Bob Bridges" <robhbrid...@gmail.com> Sent: Sunday, September 06, 2020 6:18 AM > When you care about efficiency, I'd think this would be better: > > const=4/3*3.14159E0 /* in the initialization */ > volume=const*radius**3 /* inside the loop */ > > -----Original Message----- > From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On > Behalf Of Robin Vowels > Sent: Friday, September 4, 2020 18:14 > > As for writing formulas, I prefer to follow a well-known formula, thus: > > volume = 4/3 * 3.14159 * radius**3 > > However, if I'm interested in efficiency, I'd prefer > > volume = 4 * 3.14159E0 / 3 * radius**3 ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
