On 26/12/2012 19:54, Chuck Guzis wrote:
On 12/26/2012 11:30 AM, Terry Stewart wrote:
It seems things have gone horrible, 50 years ago the code where I worked
was generally good. When I worked on a professional development team we
had coding standards that were adhered to, did code walk throughs, and
had proper documentation. These days it seems we don't have time for the
checks and balances that make code maintainable and reliable.
Sigh, it seems that nothing's changed in the last
50 or so years. Why
anyone would want to write financial code in C++ is a little beyond
me. That being said, I recall a couple of financial applications that
*were* written in APL.
You will find that APL was "widely" (and may still be) used in some
financial environments, especially Insurance as its very simple to
operate on vectors and lots of insurance and other finance involves that
type of mathematics. If you are doing mathematical modelling then C++
and objects for the elements makes sense to me.
All in all, COBOL still isn't too bad for such
stuff, no matter how
some people view it. The wonderful part is that it's just as possible
to write crappy code in COBOL as it is in IPL-V.
Making programmers write actuarial code in Cobol is just plain cruel. I
well remember a "IF ..... THEN COMPUTE ... ON SIZE ERROR ....." that
blew the compiler stack because it got to a closing ")" and didn't know
if it was part of the logical expression for the "IF" , or part of the
"COMPUTE" or a subscript....
I've long wondered if the Y2K panic did more to
clean up old COBOL
code or merely dirty it up more. I suspect the truth is a little of
both.
I think much of it was done using automated tools, and I think there was
very little structural change. Just fixing the dates. Mind you when I
started programming at an insurance company in 1976 our oldest policy
was issued in 1896, and the policies I took out a few years later were
30 year policies which matured in the current century...
--Chuck
Dave
G4UGM