larry@VLSI.JPL.NASA.GOV (07/01/88)
--
I see no shame to admitting that I once did CoBOL programming; I think
language chauvinists are ridiculous. Engineers should know that every tool
has its good and bad points. The essence of engineering is to make trade-
offs, and that's as true for the tools used as the tools produced.
Business programming initially involved fairly simple calculations on large
and regularly changing data sets, and fairly complex data (re)formatting.
CoBOL was designed for this and it does this well.
Ada's calculation facilities are adequate for this job, as is the efficiency
of most Ada compilers (though there are still some rough edges). What it
lacks to match CoBOL is a standard facility for reading and formatting
data. Since Ada was designed to be extensible, this can be added seam-
lessly and without major effort.
The more progessive business programming shops use one of several
commercially available utilities that allows programmers to "paint" I/O
screens and print-outs, specifying the prompts/descriptive legends by simply
typing them in at the correct location. Programmers then specify the type
of data to be read/written in each field, its constraints, and error
processing. Once that's done, code can be automatically produced. It
usually is not terribly good but is adequate. Already at least one public
domain utility of this sort is available in the Ada SW Repository. (I
believe its in <ADA.FORMGEN>.)
The largest and most complex business data processing is typically done with
a database management system. Specialized jobs the DBMS can't handle alone
are often done by writing short programs that make calls to the DBMS for all
but the special processing.
Today "business data processing" is diverging ever more widely from their
traditional kind of processing. Ada, with its more general capabilities,
extensibility, and support for programming-in-the large (and the
capabilities mentioned in the three previous paragraphs) is a better choice
than CoBOL--assuming business programmers and managers can overcome their
laziness and the various Ada myths.
Larry @ jpl-vlsi (VLSI.JPL.NASA.GOV)