dkFlyer

Presentation - Library


The Search for The Perfect System

(Or, The Search for Truth within the Hype)

(Note: I explained this concept to Sun Microsystems in 1994 - in actuality, this resulted in 'Java'.)

(Original writing: June, 1996.)

As a Business and Computer Consultant, the single most frequently asked question that I hear is:
'What do I feel would make up the Best and Most Cost Effective Business Computer System?'

I spent five (5) years (from 1990 to 1995)
with well over 100 Totally Different Customers
Searching for this answer - to find:
the Best Hardware,
the Best Operating System,
the Best Database,
the Best Programming Language
the Best Application Group
and
the Best Network
that would be Easy to Maintain
that would be Fast, Powerful and Scaleable
that would work correctly All of the Time
and not become Obsolete almost overnight (or, ever).

During this time, I got to see:
an extremely wide variety of businesses, business applications and business requirements.
(I took on new projects specifically to work on a totally different type of project than my others.)

During this time, I also got to see:
every facet of how all of these varied types of businesses and organizations do business,
(how they handle their decisionmaking, resourses, accounting, legal, processes, etc)
as well as the varied types of problems that these businesses faced,
not only with computerization, but with their daily operations as well.

With Hardware, I tried out Acer, Altos, Apple, Dec (PC and Mini), Dell, Gateway, Hewlett Packard (PC and Mini), IBM (PC, Mini and Mainframe), Intel, Packard Bell, Sun (PC and Mini) and Wyse.

In Operating Systems, I tried out Apple A/UX Unix and Macintosh, Berkley BSD Unix and FreeBSD Unix, Dec Vax, HP-BOS (3000), HP-UX Unix 9.xx and 10.xx, IBM AIX Unix and OS/2, Linux Unix, Microsoft Dos, Windows, Windows NT and Xenix, Minix, Novell 3.x, 4.x and Univel and Unixware Unix, SCO Unix and Sun Interactive and Solaris Unix.

With Databases, I tried out Clipper, DB2, DBase IV, Visual DBase and Paradox, Informix 3.0, SQL, 4GL and R4GL, Microsoft Access 2.0 and 7.0, FoxPro 2.6 for Dos, Mac, Unix and Windows and 5.0 for Windows and Microsoft SQL, Oracle SQL 6.xx and 7.xx and Pro-C, Progress, RBase and Sybase.

In Languages (on many different platforms and versions), I tried out Assembler, Basic and Visual Basic, C, C++ and Java / J++, Cobol, Fortran, Pascal and Delphi and PowerBuilder.

With Applications, I tried out Corel Office (WordPerfect), Lotus Smart Suite, Microsoft Office along with many, many packages, such as Aldus (Adobe) PageMaker, CorelDraw, etc.

And, I set up Client Server and Peer to Peer Networks with AppleTalk, IPX, MS-Lan, NetBEUI, TCP/IP (with NFS) and TNS (Oracle) over Arcnet, Ethernet and Serial (including SLIP, PPP and UUCP).

And, I worked on them all Extensively!

Now, I have my own solution to this question that could
Save Millions of Dollars for large and small businesses alike.


The Library


(Original writing: June, 1996.)

Library


Purchase and Build a Library of Callable Functions
with a Database Enhanced Help System
for All Applications to Use as a Programming Base.


Class

Now, each Program would begin with Thousands of Lines of Code Already Written.

Coding would be Hundreds of Times Faster and be completed for Pennies on the Dollar
allowing for Much Greater Quality Programs and Many More Additional Programs,
(with many Easily Replacing Purchased Programs and Licenses, Saving $Millions More).

Every Program would Look, Feel and Work Alike.
Education and Training Costs would be Significantly Reduced, Saving Even More $Millions.

As One Program is Improved, All Programs would Automatically Improve, too.

We could Create Anything. 'Can't Do That' would become an Obsolete Phrase.

We could Build Testing, Logging, Debugging and Version Control Right Into the Library.

We could Build Security, Who-Tracking, Data Encryption,
Anti-Virus and Anti-Hacking Routines Right Into the Library
, too.

Programs would Never Become Obsolete since we are Constantly Building.
And, since we are building Cross-Platform,
we would work on Any Operating System and Any Hardware Platform We Choose
(including Linux, which helps to make up part of the Library Core).

Data would become Interchangeable between Applications
allowing for Much Better Reporting, Auditing, Analysis, etc,
as well as allowing for a Single Entry Database
(allowing for the Creation of the Appointment to Trigger Every Event to Conclusion).

Business gets All of the Source Code for Everything.

ISO Compliance would be Built Right Into the Library
in Programming Design, Finished Programs and with Hospital Functionality, itself.


ComputerX

Development Time Lines between Languages


(Original writing: June, 1996.)

Computer-X


The Cross-Over Point between Visual Basic and C used to be about 300 Hours.
In other words, if you were going to write a Simple Program, you would use VB.
If you were going to write a Difficult Program, you would just use C.

Now, with Borland C++ Builder (or Browsers), it is more like 50 Hours or Less.

Basically, Simple Beginnings are meant Only for Simple Programs.
Serious Programming is Always Completed in Less Time and with Much Greater Results
if Started with the More Serious Languages of the C-Family.

Plus, the ComputerX Principle can be applied to Every Aspect of the Computer.
Hardware works the Same Way.
Databases work the Same Way.
Applications work the Same Way.
Even Operating Systems (or 'Especially' Operating Systems) work the Same Way.

Computer-X Mixing Languages


Furthermore, Interacting Programs,
such as using ODBC and Extra, or mixing Visual Basic with Access, SQL, Excel, etc.
cause Exponentially More Programming Costs,
Errors and Headaches which you almost Never get with the C-Family.

Language Work-Arounds cause even More Programming Costs,
Errors and Headaches which you also almost Never get with the C-Family.

Computer-X w/ 'C-Family'


Now, the Library would Turn the ComputerX 'Upside Down'.

Programming would Begin Simple and Continue Simple,
and still be Very Serious and Powerful.

Truthfully, the C-Family with the Library should have a Difficulty Rating Far Below the Chart (above).

Compare the Library with the Cobol Library on the Mainframe, which is a proven programming practice.
Except that our Library will be written in the 'C-Family' (the Mother of all modern languages), a language that has stood the test of time, has proven to be Limitless in it's capabilities, will work on any platform, has more Libraries available to it than all other languages combined and is the only choice of all major software development around the world.

Further, DLL's could still be Created from the Source Code
(rather than purchased as VBX's or OCX's)
so the Simple Languages could still be utilized, if desired
(although I can't imagine why any would).


The Perfect System


(Original writing: June, 1996.)

Business PC's should work just like 'Mainframe Terminals',
except that they can Do More Locally.

First and foremost comes Hardware.
You should not buy an 'off the shelf' computer.
You must build one.
You must buy 100% compatible hardware that is 100% quality.
(This does not necessarily mean biggest and best.)
For a few dollars more, you save thousands in headaches.
You can build only what you need.
And, you can get it fixed immediately.

The following decisions requires the understanding of the costs involved in educating and training
all of the necessary employees in the use of new and updated software.
The greater the similarities between all of the software packages on the system,
the greater the opportunity to reduce this cost.

From this comes the choice of Operating System and Software.

The primary uses of software include databases, word processing,
spreadsheets, desktop publishing and communications.

Virtually every commercial software package on the market can be purchased in quite similar form with the source code.
Even some of the Operating Systems can be purchased (almost free) with the source code.

So, why would any business even think of buying a 'Hundred' or a 'Thousand' -Site License
when you can build your own - Better - for Free?

If you are purchasing five (5) or more of a specific package, you should be purchasing the source code, instead. Then, once you own the source code, you are legally allowed to put your own new program on any and all computers that you choose, saving potentially millions of dollars in purchases, upgrades and education and training costs, as well.

You could choose to add sound, images, video, and on and on. Any changes could be either purchased or programmed and you could create several modified versions for different employees and for faster or slower computers, recovering now obsolete hardware.

You would then also have the flexibility of creating your own
integrated database, word processor and spreadsheet.
You could even integrate them right into your own operating system,
eliminating software communication problems.

I am now using a version of Unix (Linux) that comes with the source code.
I've had the source for X-Windows, TCP/IP, the Fox-Style Database, libraries to other databases, the source for images, sound, fonts, printers, communications, faxes, voice, etc, etc, etc.
This source is all purchased in ANSI C/C++, of course because this is the fastest, most powerful and most portable language.

And, I am currently working on a Browser / Server (and OS) that supports
HTML3, VRML2, SGML, Java and CSCW (GroupWare).
It will run on a Super Server.
Or, on a 386 (or less).
Or, an NC (Network Computer).
Or, PC104 (Embedded Controller).
It will have Direct Database Connectivity.
And, will be Office Capable (Word Processing, Drawing, etc).
Because it is GroupWare, it will handle NetMeetings, etc.
I am going to use it in our Wearable Computer.
I am going to add to it our 'PIC' Application,
including our Intelligent, Decision-Making Data Analysis Tool.
And, I am going to use it to Control Machinery, too.

And, because I still (temporarily) need to use a few commercial packages
(such as Corel Draw for Windows 95),
I have set up the network using NFS and the Client Server to allow Windows 95 and OS/2
to interface with my own system (the second version of UniDos).

Designing and building your own networked computer system, based on current source code available on the market, is by far the cheapest and most reliable method that you could choose today.

You could get out of having to learn and use multiple software packages just because one package has one or two different features than another.

You could purchase the source code to create files for, or to link to major commercial software packages such as Excel, Lotus 123, Ami Pro, Word Perfect, Word, etc
(for businesses outside of your own organization, of course).

You could create features specific to your own business needs.

You could utilize tremendously cheaper hardware.

You could get a bug fixed your own self.

You could actually have control.

You could even choose to do something absolutely unheard of:

You could choose to make your own computer system - simple.


Turning 'The Presentation' into 'The Plan'


(Original writing: June, 1996.)

Buildables

Item Comment
We need Numbers on the Failure of the 'Name Brand' PC's. From our meeting, it was more than 33% (with half of that right 'out of the box') within only 30 days.
We need Costs of the 'Name Brand' PC's to compare with our own. Our Buildables could also Re-Use the 3Com Ethernet Cards currently in the 'Name Brands'.
We need Facts and Opinions on the Ease of Setting Up and Fixing our Buildables compared with the Difficulties of the 'Name Brand' PC's. Including the Difficulties in using So Many Different Ethernet Cards, Video Cards, CD-Roms, etc... as well as the Fact that it is Our Buildables that are the Real, Quality PC's, while Theirs are the Generics.

Library

What was the Cost of our Custom Software Development for the Last One to Five Years?
What is in our Budget for Custom Software Development for the Next One to Five Years?

What was the Cost of our Software Purchases (including Licensing)?
What is in our Budget for Software Purchases (including Licensing)?

What was the Cost of our Software Training (both Custom and Purchased)?
(Setting up for Training, the Instructors, the Trainees (including Doctors and Nurses), the Occurances of Training, Programmers to Learn our Custom Source Code, etc...)
What is in our Budget for Software Training (both Custom and Purchased)?

What is the Budget for 'Software and Development' (for comparison)?
Are we an Alternative?
Or, a possibly an extremely inexpensive Hedge if 'Software and Development' doesn't work?
How do all of these costs compare with the Total Costs of the Library?

Author: Dennis Kennedy
Copyright © 1996 dkFlyer
If you have any questions, email me.

Hosted by www.Geocities.ws

1