1995-10-12 - Re: [NOISE] was Re: java security concerns

Header Data

From: “Perry E. Metzger” <perry@piermont.com>
To: madden@mpi-sb.mpg.de>)
Message Hash: f28fd44b34df89abdd69df7885f0e87dde854ea2dc8e453568c4430f5b2867dc
Message ID: <199510121346.JAA17394@jekyll.piermont.com>
Reply To: <199510111335.JAA17959@panix.com>
UTC Datetime: 1995-10-12 13:48:17 UTC
Raw Date: Thu, 12 Oct 95 06:48:17 PDT

Raw message

From: "Perry E. Metzger" <perry@piermont.com>
Date: Thu, 12 Oct 95 06:48:17 PDT
To: madden@mpi-sb.mpg.de>)
Subject: Re: [NOISE] was Re: java security concerns
In-Reply-To: <199510111335.JAA17959@panix.com>
Message-ID: <199510121346.JAA17394@jekyll.piermont.com>
MIME-Version: 1.0
Content-Type: text/plain



Peter Madden) (by way of Duncan Frissell <madden@mpi-sb.mpg.de> writes:
> The real problems lie with specifying the
> program/problem correctly in the first place (so-called specifications
> capture), and with automatic program *synthesis* from specifications
> (which, in mathematical theorem proving terms, presents the problem of
> creating existential objects, as opposed to just verifying that they
> do the right job).

Bugs in specifications are just as easy for humans to produce as bugs
in implementations, and unfortunately there is no way for our machines
to psychically intuit what it was we wanted to specify that they do
any more than we can make them intuit what it was we wanted them to
do.

Remember, by the way, that in some sense a high level programming
language *is* a specification language. Authomatic synthesis from
"specifications" is just a higher level of programming language, with
all that entails.

Perry





Thread