Path: utzoo!attcan!uunet!bionet!arcc.afrc.ac.uk!ODONNELL
From: ODONNELL@arcc.afrc.ac.uk
Newsgroups: bionet.software
Subject: Software publishing
Message-ID: <9002131952.AA00244@net.bio.net>
Date: 13 Feb 90 17:27:00 GMT
Sender: daemon@genbank.BIO.NET
Lines: 27

Re : Software publishing

A basically good idea to help other scientists avoid "reinventing the wheel"
for those special tasks. But we should be careful of adding to the computer
equivalent of 'journal overload', where the junk obscures the good quality.

Much of the 'academic' software, even that published in the journals fails on
the following counts:
    1) Poor presentation
    2) Lack of error trapping and input-checking.

This is understandable, if only because the academics who write the software
have not developed it for public use. But the fact that a scientist finds
it useful does not make it 'good software'.

I think the selection procedure would be very important to screen out poor
code and poorly implemented algorithms. This would be quite a demanding task
for someone who should be aware of both the theoretical basis of the algorithm
and good coding practice.

*************************************************************************
Cary O'Donnell			Tel: (in UK) 0582 762271 ext 226
AFRC Computing Centre		     (+44 for UK) 582 762271 ext 226
West Common			Fax: (+44) 582 761710
Harpenden
Herts AL5 2JE			email: ODONNELL@UK.AC.AFRC.ARCC
U.K.				(Molecular biology support at AFRCCC)