Subject: Re: standards & FS
From: "Stephen J. Turnbull" <>
Date: Fri, 28 May 1999 19:17:03 +0900 (JST)

>>>>> "Ian" == Ian Lance Taylor <> writes:

    Ian> If you do have the source code, you don't need an
    Ian> independent definition, because the source code itself is a
    Ian> precise definition of the interface.

No.  It's precise on what's available, it is not precise about what's

    Ian> So I think standards are less important for the free
    Ian> software community: we have the source code.


    Ian> However, an implementation defined standard is slightly
    Ian> inaccurate in ways that are hard to understand (because the
    Ian> implementation has bugs),

Irrelevant; any design can have bugs.

    Ian> it is liable to change unexpectedly (because the maintainer
    Ian> may see a better way of doing things),

Sometimes the maintainer already knows how to do this, or may want to
produce plug'n'play implementations tuned to different usages.  "All
symbols not explicitly delegated to the interface are reserved to the
implementation."  The Founding Fathers got _that_ one right.  :-)

    Ian> and it is difficult to work with (because it's easier to read
    Ian> a natural language text than to dive into a big program and
    Ian> figure out what it does).

It's not natural language; we use BNF or yacc pretty easily.  Natural
language helps, of course, but the real benefit to a separate standard 
is that you don't have to cut out the irrelevant yourself.

    Ian> So I think the free software community still needs standards.

Agree.  Standards are a form of design.  The FS community needs to be
doing design in general, and standards in particular, at least if we
want to implement business models beyond the commoditization and service
aspects of software.

University of Tsukuba                Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Institute of Policy and Planning Sciences       Tel/fax: +81 (298) 53-5091
What are those two straight lines for?  "Free software rules."