Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!asuvax!noao!arizona!gudeman
From: gudeman@cs.arizona.edu (David Gudeman)
Newsgroups: comp.lang.misc
Subject: Re: Formal definitions (Re: ada-c++ productivity)
Message-ID: <1954@optima.cs.arizona.edu>
Date: 15 Apr 91 23:47:05 GMT
Sender: news@cs.arizona.edu
Lines: 31

In article  <50835@nigel.ee.udel.edu> Darren New writes:
]
]>A _formal_
]>specification can be as wrong as a program can.  
]
]Usually not, because the specification is at a higher level and closer
]to the way a person thinks than an implementation.

That is an argument for higher-level programming languages (with
dynamic typing, for instance), not an argument for adding another
language (the specification language) to the process of implementing
software.  If your specification is so great, why not make that your
programming language?

]... experience has again and again shown that formal
]specifications tend to have fewer bugs than informal specifications.

I'll bet that is heavily dependent on the application.

]...  Every time I've heard of that
]somebody has taken (say) an ISO network protocol specification and
]respecified it formally, bugs have been found...

When I try to implement an informal specification in any programming
language I usually find "bugs" in the specification.  No win for
formal specs there either, it is just the process of going into
details that finds the problems.
--
					David Gudeman
gudeman@cs.arizona.edu
noao!arizona!gudeman