>I've never believed, and will never believe that rewriting a full business application will ever work, unless you suddenly can double or even quadruple your dev. resources, just because of the innumerable failure experiences documented.
>
>
>>Thierry,

One of the many nice things about advancing in years is that you get to witness cycles come and go.

First I saw the mainframe COBOL cycle.

I was fortunate to be working in a large DP shop when the IBM 360 was announced.
That announcement rocked the DP world at that time and basically blew away IBM's competitors.
Mainframe COBOL running on 360 architecture became a standard that would dominate DP for decades. I personally rewrote several machine language programs for that platform as did thousands of programmers around the world.
The justification was simple, one COBOL programmer with a lot less training would out produce five programmers writing machine language and a human being could read the code.
Simultaneously, the IBM 1401 replaced tens of thousands of tabulating machines around the world.
That required writing IBM 1401 code to replace the wired boards of the tabulators, but the economics were so compelling that you didn't have to think about it.. you just did it.

That cycle started to wane when LAN's started to become efficient during the late 1980's



I adopted FoxPro during the late 1980's because I needed a tool that would allow me to replace aging and comparatively expensive mainframe data processing systems (back then, we called it data processing, which is really what it is) with LAN's.
The economics were compelling- for an outlay of under $100K, a client could replace a system costing over $150K/yr to operate.
I built a lucrative business on that model.
My group replaced mainframe COBOL systems with Foxpro systems.
We shut down several large data centers in NYC and replaced them with LAN's with huge savings in operating costs and increased productivity for the end user.
When the Y2K came along we couldn't meet the demand and had to turn away many large lucrative contracts.
Basically, that cycle started to wane after the Y2K and now VFP and its counterparts have become the legacy apps.

So.. not only did rewriting a full business application work, it was a huge success in every case that we took on.
Have you heard of anyone saying that we should go back to mainframe COBOL lately?
Or.. have you heard of anyone lamenting the loss of tabulating machines?

Today there is no clear justification for rewriting most VFP systems because the alternatives are not more efficient economically.
Most of the justifications that I hear are intangible, and my first boss, when I tried to justify an expense with intangible benefits, asked me:

"How would you feel if I put those intangibles in your paycheck, instead of dollars?"

So.. we are in the same positions those COBOL shops were in after the PC had arrived but before Foxpro and an efficient LAN came along.
We know that there's probably a better way of doing what we're doing, but can't find a compelling case for any one choice yet.

I switched to .NET because of the market demand for it.
In 1984, I'd have written COBOL for the same reason, while knowing full well that something better, but yet unknown, was on the horizon.