Subject: Re: dynamic linker
To: NetBSD User's Discussion List <netbsd-users@NetBSD.ORG>
From: Greg A. Woods <woods@weird.com>
List: netbsd-users
Date: 03/19/2002 16:20:53
[ On Tuesday, March 19, 2002 at 19:34:58 (+0000), Sean Witham wrote: ]
> Subject: Re: dynamic linker
>
> Has anyone done a good white paper on the pitfuls of 
> dynamic linking ?

I've read some academic papers about the pitfalls of dynamic linking,
and I know at least a few other software engineers who've hurled insults
at every dynamic linking scheme they've ever run across.

SunSoft's Chief Scientist, Rob Gingell, wrote a "white paper" giving
their position on static vs. dynamic linking options.  Much of what he
says is true, though it's blatantly slanted in the direction you'd
expect from a vendor of proprietary software.  Some of what he says is
very misleading.  The rest is, at least in my opinion, a house of rather
wobbly cards.  Oddly depsite all the hand waving about tuning and
optimisation I don't think Sun have done anywhere near as much on this
front as is done in Darwin.
<URL:http://www.sun.com/sun-on-net/itworld/UIR960201perf.html>

John Levine's "Linkers & Loaders" book (pub. Morgan-Kauffman, Oct. 1999,
ISBN 1-55860-496-0) contains two relevant chapters, #9, "Shared
Libraries", and #10, "Dyamic Linking and Loading."  Unfortunately the
main points in the opening paragraph of chapter 10, which address the
"benefits [of dynamic linking] that are hard to get otherwise" are all
wrong (at least in many common circumstances, though not always for
obvious reasons).  The disadvantages he does mention though are very
real.  (The rest of the chapter though is very informative!)
<URL:http://www.iecc.com/linker/linker10.html>

-- 
								Greg A. Woods

+1 416 218-0098;  <gwoods@acm.org>;  <g.a.woods@ieee.org>;  <woods@robohack.ca>
Planix, Inc. <woods@planix.com>; VE3TCP; Secrets of the Weird <woods@weird.com>