Return to BSD News archive
Newsgroups: comp.os.386bsd.development
Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!agate!howland.reston.ans.net!gatech!news.byu.edu!cwis.isu.edu!fcom.cc.utah.edu!cs.weber.edu!terry
From: terry@cs.weber.edu (A Wizard of Earth C)
Subject: Re: SHARED LIBRARIES - THE END
Message-ID: <1993May24.225014.23425@fcom.cc.utah.edu>
Sender: news@fcom.cc.utah.edu
Organization: Weber State University (Ogden, UT)
References: <PC123.93May22195506@bootes.cus.cam.ac.uk> <1993May23.003623.24102@fcom.cc.utah.edu> <1tr05o$qaa@terminator.rs.itd.umich.edu>
Date: Mon, 24 May 93 22:50:14 GMT
Lines: 102
In article <1tr05o$qaa@terminator.rs.itd.umich.edu> pauls@css.itd.umich.edu (Paul Southworth) writes:
>In article <1993May23.003623.24102@fcom.cc.utah.edu> terry@cs.weber.edu (A Wizard of Earth C) writes:
>>My personal misgivings on the issue are based on me being, basically,
>>in competition with Pete for the shared library consumers, and the
>>timing with regard to the 0.2 release.
>
>Um, how can there be competition when 386BSD is not a commodity? You all
>are producing freely distributable code, right? As long as there is
>interest in both packages, why should there be only one? I mean, apart
>from your respective egos, in what sense can this competition exist?
>Please clarify. I mean, it would be neat to have the two of you work on
>one package, but I don't consider it a great loss that you choose to
>work separately -- it's your code after all.
Pete's code is about to be dubbed "official" in Bill's 0.2 release. I
think it's a good interim soloution (hell, I'm using it!), but it seems
to me that it will be extermely difficult to displace it once it becomes
part of a distribution. Pete's code will be the defacto standard because
of it's release as part of the base release.
There is an implicit assumption that Pete and I were aware of each other's
work, or that Pete was part of the working group discussions on ref, or
that we don't both have a million things going at once.
You also seem to have missed the gist of my message, which was "It's a shame
that people have flamed Pete into hiding; he's done a good job, and they've
only robbed themselves of the future benefits of his work".
Before Pete's code came out, I used Joerg's code (which Pete's code updates).
But there are some severe restrictions placed on the directions future
code can go based on SVR3 style shared libraries (eg: Joerg's approach).
A lot of this was discussed on the "chat" meetings (some of which involved
Bill) on the old "ref" machine; they were chaired by Julian.
"Compete" is in terms of "people will use one or they will use the other of
two implementations", not on the basis of "OK, who gets this dollar?".
I think as long as no implementation was "dubbed", then there wasn't a
problem; it's certainly wrong to criticize Pete for the code he has
provided, or to criticize the approach taken i the code (Joerg's approach);
but it's right to criticize the cannonization of that code by Bill, since
the code is useful for a user, but restricts the direction that future
developement can take. It is in conflict with the stated goals of 386BSD
to include the code.
In the same fashion, 386BSD "competes" with Linux. It's a popularity
contest, and it boils down to "My baby is perfect, but yours looks like a
miniture elephant-seal". The GNU vs. Berkeley License issue is only
secondary, and is more akin to Plastic vs. cloth diapers.
Bill has been getting beaten up about a lot of things: closed developement,
time between releases, revision numbers he should use, the difference
between doing research and providing a research platform (the whole C++
thing), and not having shared libraries. He can do something about
shared libraries Pete's code *now*, without opening himself up to more
of the other arguments (in particular, time between releases). There
is great end-user pressure for him to opt for this, and it satisfies the
criteria for a counter example to some of the arguments he's been hearing.
If the issue were to be resolved without cannonizing one approach in the
research platform until after the research is done, then I wouldn't have
a problem. I dislike the pressures that are forcing Bill into the
decision to support one approach over another *on any issue* while there
is still research taking place. I view the official 386BSD releases as
codifying one piece of research over another; as far as shared libraries
are concerned, this shouldn't be done yet.
Unlike the SCSI drivers, for which there is a clear rationale for Julian's
work, there is no clear rationale for shared library approaches other
than "one's available". Also, unlike SCSI drivers, the machines which
can run 386BSD are not limited by the existance or non-existance of shared
libraries. The arguments for "memory" and "disk space" don't hold water,
since Bill's release is to provide a research platform, not an end user
system (that's NetBSD's goal, and they should feel free to repackage 386BSD
any way they want for end user consumption, including picking and using a
shared library package in their distribution). You can't install 386BSD
on some machines without the SCSI drivers; you *can* install, in all cases,
without the shared libraries; at the very least, a sufficient minimal set
can be installed so that shared libraries can then be installed if the
machine is capable of supporting more than a minimal set without them. If
not, the machine is incable of supporting a minimal set -- again the argument
for shared libraries in the 0.2 release is lacking.
Pete and Joerg have done some fine work, and if you are an end user or if
the direction of research you pursue isn't hampered by their implementation,
you could easily choose to use those libraries; I currently use it for one
of the several 386BSD machines I have up. But if what you are talking about
is a research platform ,there should be as few restrictions on usage as
possible. The goal of 386BSD is to provide a research platform, not a user
platform, and it should not be subverted lightly.
"Let the Flames begin!" (my keyboard is missing italics)
Terry Lambert
terry@icarus.weber.edu
---
Any opinions in this posting are my own and not those of my present
or previous employers.
--
-------------------------------------------------------------------------------
"I have an 8 user poetic license" - me