From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me
X-Spam-Level: 
X-Spam-Status: No, score=0.6 required=5.0 tests=BAYES_40,INVALID_MSGID
	autolearn=no autolearn_force=no version=3.4.4
X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: 103376,69d15fa7a1d094e1,start
X-Google-Attributes: gid103376,public
From: tmoran@bix.com
Subject: Re: Info About Real Ada 95 Projects
Date: 1996/08/04
Message-ID: <4u19qf$g4s@news2.delphi.com>#1/1
X-Deja-AN: 171932576
organization: Delphi Internet Services Corporation
newsgroups: comp.lang.ada
Date: 1996-08-04T00:00:00+00:00
List-Id: <comp.lang.ada>


Just to add a non-militiary, but small, project:
  Two years ago I used RR's beta Ada 95 for a one-shot project to
clean historical financial data for a data retailer/investment counsel
company.  Their internal database of stock and commodity data had gaps
and errors.  They bought a database from a well know economic
consulting firm - which also had gaps and error, but hopefull
different ones.  My program merged the two, including detecting and
correcting for stock splits.  Since true data is sometimes difficult
to distinguish from erroneous, the program ran one task that ran
through the CDROMs and hard disks of data, while another flashed
alternating charts of two versions of a particularly peculiar
financial history so a human expert could accept one or note for
further research.
  Since the owner of the firm was a self-taught C++ programmer, and
considered himself as expert at that as at investments, it seemed
unnecessary to discuss my choice of language with him.  Since it was a
one shot probject resulting in a new, improved, data base, there was
no question of future code maintenance.  It's thus one of those
'hidden Ada' projects.
  This project did not use any of the 'big' OO, tasking, or child
library features of Ada 95.  It did use little things like 'use type',
modular types, etc.  One of the databases used scaled values which
were just right for 'fixed'.  Subtypes and exceptions of course helped
catch the truly bizarre.  While demoing an early version it crashed.
Using the exception traceback I was able to find the kind of weird
data that caused the problem, put in explicit handling, recompile, and
return to the demo in under 10 minutes.  The client was pleased.  I
still see large ads in investment magazines for his 'high quality data'.
Tom Moran