Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!yale!cs.yale.edu!spolsky-joel
From: spolsky-joel@cs.yale.edu (Joel Spolsky)
Newsgroups: comp.windows.ms.programmer
Subject: Re: Add-ons
Message-ID: <27097@cs.yale.edu>
Date: 6 Nov 90 21:16:21 GMT
References: <90309.111451NIBMSCM@NDSUVM1.BITNET> <27085@cs.yale.edu> <1990Nov6.180958.17256@world.std.com>
Sender: news@cs.yale.edu
Organization: Yale University Computer Science Dept., New Haven, CT 06520-2158
Lines: 19
Nntp-Posting-Host: zoo-gw.cs.yale.edu
Originator: spolsky@suned.CS.Yale.Edu

In article <1990Nov6.180958.17256@world.std.com> goodearl@world.std.com (Robert Goodearl) writes:
[ about MSC Library Source ]
>I think the sources would be very useful if you needed to better understand
>the behavior of a library routine, or to make romable code.

In fact, reading over some of the assembler code on those disks, I was
truly impressed by what hoops MS jumps through to squeeze more speed
out of their libraries. For example, string/memory copy functions use
the Intel memmov (REP MOV) instruction, but they also make sure to
word align for improved performance -- these libraries contain an
impressive amount of micro optimization, enough to scare me out of
every wanting to "roll my own".

But prepackaged libraries are not always the fastest, I once looked at
the code in the MPW (Apple Macintosh) C libraries and was unimpressed;
I wrote a strcpy which outperformed their library version.

Joel Spolsky
spolsky@cs.yale.edu                                     Silence = Death