Hey guys

While you are all waffling on like a bunch of over paid consultants
trying to justify your fees the end of 1999 approaches relentlessly.

If we are going to ensure that we do not have the much talked about
shambles/disaster/whatever we need to prove that the African
practical approach to life works. I dont mind if you are from the
north or the south. Give or take three hours the problem will hit us
all at the same time and we need to get on with it.

In my simple mind the problem is a seven layer technical issue which
must be addressed by those people who have the ability to solve it.

It really doesnt matter if there is "management buy in", "user
ownership", or what ever.

As workers with IT the problem is ours. If we get it right no one
will notice, and thats good, while if we get it wrong a) we will get
the blame and b) we will have to fix it any way.

By now we must all understand that there is a problem that will effect
virtually all computers and equipment that uses technology and should
be living with a simple plan that says:-

    1.  what have I got?
    2.  what does it do?
    3.  will it do it properly when the date changes?
    4.  how do I test it?
    5.  how do I set the correct priority?
    6.  can I fix it before or after?
    7.  what will be the impact of it failing?
    8.  what will I do if it doesnt work?
    9.  how will I handle those failures that I dont anticipate?

Its that simple and that complex!

The seven layers, up from four, for those who dont know are:-

    1.  hardware    -   on pcs test with free ware software
    2.  micro code  -   on pcs test with free ware software
                        fix with patches from suppliers or other
                        sources if you can get them !!!
    3.  programming languages
                    -   hard to test
                    -   need input from manufactures
                    -   likely to be the hidden Achilles heel for many
                        and may actually be the time bomb we talk
                        about (imho)
    4.  application programs
                    -   check by inspection
                    -   test with y2 data
    6.  application data
                    -   hard to check except by programmatic scan
    7.  interface with other systems
                    -   problems here of data format
                    -   problem with transfer mechanism, eg Internet

This model, which I use widely in publicising the problem, appears to
apply to all computers be they, mainframe, mid range, pcs, embeded
control systems, network components, etc.

Much of the retoric that was aired recently has been aimed at only
one or two components of the model or of the plan.

The difficulty in understanding the bigger picture, rather than just
ones own problem, is what causes the kind of generalisations we have

Do not be fooled. The problem is does exist and it is ours. Solutions
ranging from the very simple to the very complex do exist but there
is no "Silver bullet" out there just waiting for large lumps of
dollars, sterling or Euros, there is only sweat. If there was a money
solution various US states and large commercial enterprises would not
have frozen all development work to concentrate on the solution.

Good luck and work hard but plan clever. We are past the time for
academic debate its now time to get in there and fix.


      CITTE 98 at url

Subscribe to the IT Digest, an information resource from Wits Univ.
Send e-Mail to [log in to unmask]  with SUBSCRIBE ITDIGEST
and {your_user_id} in the body followed by END on the next line.
Henry C Watermeyer                         'Phone +27-11-716-3260/8000
Director - Computer 7 Network services     Fax    +27-11-339-1225
University of the Witwatersrand, Johannesburg
P/Bag 3, Wits 2050, South Africa           mobile +27-(0)82-800-8862
//      //