Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!jarthur!uunet!kddlab!cs.titech!titccy.cc.titech!necom830!mohta
From: mohta@necom830.cc.titech.ac.jp (Masataka Ohta)
Newsgroups: comp.unix.wizards
Subject: Re: Shared libraries (was Re: Window system bashing (was Re: X11 bashing))
Message-ID: <154@titccy.cc.titech.ac.jp>
Date: 30 Apr 91 04:29:54 GMT
References: <1991Apr25.175847.21613@kithrup.COM> <BZS.91Apr28000804@world.std.com> <73761@brunix.UUCP> <1991Apr29.025800.3108@kithrup.COM>
Sender: news@titccy.cc.titech.ac.jp
Organization: Tokyo Institute of Technology
Lines: 29

In article <1991Apr29.025800.3108@kithrup.COM>
	sef@kithrup.COM (Sean Eric Fagan) writes:

>On a '386 running SCO UNIX (or xenix, for that matter), a call to a shared
>library function one jump in addition to the normal call.  Thus, there is a
>*very* slight performance hit.

Branch is the most ineffecient instruction on today's highly pipelined
architecture.

>I don't necessarily agree that most of libc should be shared.

Didn't you claimed sharing of libc is good because of UUCP and MMDF?

>libX, on the
>other hand, as well as the toolkit libraries, would be a *very* good thing
>to share, for many reasons;

Well, the following equation holds:

	uni_X + X_window = multi_X

>the increase in execution time is offset far
>more than enough by the savings in paging, physical memory, and disk space
>used.

The best solution is not to have a complex window system.

						Masataka Ohta