Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uunet!mcsun!ukc!ox-prg!client40.comlab!ptb
From: ptb@prg.ox.ac.uk (Peter Breuer)
Newsgroups: comp.software-eng
Subject: Re: Summary: Restructuring and Reverse Engineering.
Summary: Successful Reverse Engineering tools exist. The `summary' is off-mark.
Keywords: Reverse Engineering
Message-ID: <1789@culhua.prg.ox.ac.uk>
Date: 27 May 91 12:18:15 GMT
References: <1991May25.095806.3732@weyrich.UUCP>
Sender: news@prg.ox.ac.uk
Followup-To: comp.software-eng
Organization: Oxford University Computer Laboratory (Programming Research Group), England, &  Cambridge University Engineering Department, England
Lines: 46

In article <1991May25.095806.3732@weyrich.UUCP> orville%weyrich@uunet.uu.net
writes:

> MY CONCLUSIONS:$(I found no comments on successful reverse engineering
> products beyond%(simple cross-reference generators and flow-chart
> generators.'(Restructur ing is widely done; [...]

The bit of Kevin's response that Orville posted

>>We are interested in applying machine-learning techniques to determine
>>useful restructuring strategies, but not much work has yet been done by
>>us on this. Tools that transform code to the functional language, and
>>support the user in re-writing this code, have been developed in Prolog.

doesn't seem to say so, but our (mine and Kevin's) research work within the
REDO project (Esprit project 2487) is aimed at producing **abstractions**.
Our research has always been successful and ERGO we have successful reverse
engineering tools which produce abstractions from the code. That is somewhat
different from `simple cross-reference generators and flow-chart generators'
(fume! fume! gnash! gnash!).

I suppose the claim will be that by 'product' was meant `commercially available
product'. Still, send me money and I'll see what I can do. Since I have to
spell it out ....

1) We produce exact (but readable, because they have been `optimized'
automatically for readability) translations in a first-order functional
programming language *first*. I suppose a call-diagram view of this could be
called a `flow-graph' (snort!).
2) Human beings then interact with the reverse engineering tool and
progressively organise the mathematical descriptions of the functions into
boxes which end up as objects in an object-oriented description, whilst the
application code becomes progressively impoverished of all but calls to object
methods.

The *final* abstraction is in the language called Z++. This is an o-o
extension of the specification language Z. For references, mail me (or
read the references quoted in the original posting). Moreover, the following
either has appeared or will appear SOON:

Breuer~P.T., Lano~K.C.
Creating Specifications from Code.
Journal of Software Maintenance, to appear 1991.

--------------------------------------------------------------------------
Peter T. Breuer  | OUCL: ptb@uk.ac.ox.prg | CUED: ptb@uk.ac.cam.eng