Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!nrl-cmf!ames!vsi1!wyse!mips!mash From: mash@mips.COM (John Mashey) Newsgroups: comp.arch Subject: Re: String lengths Message-ID: <13249@winchester.mips.COM> Date: 14 Feb 89 15:52:07 GMT References: <8876@alice.UUCP> <765@atanasoff.cs.iastate.edu> <637@m3.mfci.UUCP> Reply-To: mash@mips.COM (John Mashey) Organization: MIPS Computer Systems, Sunnyvale, CA Lines: 22 In article <637@m3.mfci.UUCP> colwell@mfci.UUCP (Robert Colwell) writes: .... >Dennis's comment *could* have been circular, but I don't think it >was. After all, the Unix OS has lots of places where exceptionally >poor string handling would be obvious very quickly, and there are >several known installations of this OS nowadays...On the other hand, >there's Dhrystone -- if your string handling is poor, your >Dhrystone number may well be pathetic. And Dhrystone is supposed >to be a systems code benchmark (at least to this level of fidelity). As has been discussed before in this group, Dhrystone's str* behavior seems to differ a bit from "more typical" C programs. Everything I've ever seen from analyzing C programs backs up what Dennis says. When we did the first MIPS UNIX port, I wrote all of the str* routines in assembler; we threw most of them out in favor of the portable C programs, because the testing time to find the (must have been at least) one bug in the lot wouldn't have been worth it, according to the program statistics we saw. -- -john mashey DISCLAIMER: UUCP: {ames,decwrl,prls,pyramid}!mips!mash OR mash@mips.com DDD: 408-991-0253 or 408-720-1700, x253 USPS: MIPS Computer Systems, 930 E. Arques, Sunnyvale, CA 94086