[Info-vax] Programming languages on VMS

[Arne Vajhøj]

On 1/24/2018 2:51 PM, Bill Gunshannon wrote:
> On 01/24/2018 02:20 PM, Arne Vajhøj wrote:
>> On 1/24/2018 10:45 AM, Bill Gunshannon wrote:
>>> But for much of the real work that drives business (like processing
>>> credit card transactions or computing actuary tables languages like
>>> COBOL and Fortran are still really the best choice.
>>
>> I am highly skeptical about that.
>>
>> Credit card transaction processing is traditionally done in Cobol, but
>> I can not see any particular characteristics in Cobol that many newer
>> languages does not have. I suspect that the main reason for Cobol's
>> dominance in this market is that Cobol was the best language when the
>> code was original written and that rewriting is considered too
>> risky/expensive/interrupting.
>>
>> Fortran has never been widely used for business processing.
> 
> Never said it was.  But things like Actuary Tables are  not what most
> people think of when they think business.  But is certainly is to an
> Insurance Company.  How about The Census Bureau? How would you class
> them?  They still use Fortran and they still write new programs in it.

I doubt that Fortran is an optimal choice for what Census Bureau does.
I would expect Cobol or a bunch of newer languages to be better for
that.


>>>                                                       The only thing
>>> driving the move away from them is academia's decision to drive the bus
>>> off a cliff rather than preparing students for entry into the IT world
>>> (their actual job!!) by not only not teaching the requisite languages
>>> but trying to sway students into believing the languages are dead and
>>> totally unused.
>>
>> Or maybe programming languages has also improved like most other
>> areas in IT.
> 
> As have COBOL and Fortran.

In fact both Cobol and Fortran standards has evolved a lot.

But neither implementations nor users has taken on the new
standards.

Fortran is probably doing slightly better than Cobol in this
regard.


>                         To continue to meet the needs of a specific
> domain.  So why then would you use a generic language?  

>                                                                    If
> you have a language available that was specifically designed for your
> task why would you choose to use one that wasn't specifically designed
> for any task?

I can follow that argument for Fortran. It has some characteristics
(multi dimensional array handling, the way one can access part of array,
complex number support, mathematical functions support etc.) that makes
it still unique for a certain types of application - not a huge market,
but it exists.

I find it much harder to see similar argument for Cobol. Lots of
newer languages have targeted its area. And I do not see any
unique characteristics (well - no relevant unique characteristics -
Cobol certainly have some unique language characteristics) that
can make it an obvious choice.

Arne