[Info-vax] The (now lost) future of Alpha.

[Arne Vajhøj]

On 8/3/2018 6:57 PM, Bill Gunshannon wrote:
> On 08/03/2018 06:28 PM, Arne Vajhøj wrote:
>> On 8/3/2018 8:32 AM, Bill Gunshannon wrote:
>>> On 08/02/2018 10:03 PM, Arne Vajhøj wrote:
>>>> Embedded is not really me, but I would expect the role
>>>> of C to decline as many embedded systems get much more
>>>> CPU and memory.

>>> We heard that when Ada came out.  While a lot of
>>> embedded work moved to Ada C remained a mainstay.
>>> I expect the same will be true for other languages.
>>> As the amount of embedded increases other languages
>>> may make an appearance, but not supplanting C, just
>>> in addition to it.
>>
>> C will certainly continue to be used.
>>
>> And to clarify - by "role of C to decline" I only
>> meant the percentage of embedded programs in C not the
>> number of embedded programs in C.
>>
>> But embedded with multi core SOC's and hundreds
>> of MB of RAM gives more options.
>>
>> C, C++ or Ada still provide easy HW access and
>> good real time characteristics, so they will
>> not go away.
> 
> But this is the same logic used to "prove" the decline of
> COBOL.  There is as much and likely considerably more
> COBOL today than in the 90's and yet because other languages
> are being used for mostly time wasters (you know, those apps
> on your phone) the fact that COBOL is still used for some of
> the most important aspects of business is ignored and we
> are told COBOL is dead.

Not really.

I expect the amount of COBOL code to have declined slightly.

But more importantly the amount of COBOL work has declined
a lot.

I don't think neither is the case for embedded C.

That the amount of COBOL work has imploded is very visible
from the lack of demand of COBOL skills. A lot of COBOL
programmers are retiring. A lot of COBOL programmers
are either moving to other languages or moving to
non-programming work. The influx of new COBOL developers
are very small. But still there are no long lines
of recruiters frantically lined up to try and persuade
those with COBOL skills to work for them. That
means that the the demand has decreased a lot.

Every few years there come an article in some newspaper
claiming that COBOL skills will become gold as all the
old COBOL programmers retire. But it has not happened.
Demand seems top as fast or maybe even faster than
supply.

I believe that two of the areas where COBOL code
has actually been replaced are:
* reporting - instead of having a COBOL program
   pull data out of RDBMS/ISAM file and generate
   a report, then then the data just got moved to
   a DWH and a drop and drag reporting tool is used
   to produce reports
* ERP/CRM systems - instead of bespoke COBOL applications
   then companies buy SAP, Oracle, MS Dynamics and
   customize those (and the customization is done in
   ABAP, Java, X++ etc. not COBOL).

There are still a lot of core COBOL applications
running in financial industry, government etc.. And
there is little appetite for replacing them - it will
cost a fortune and there will be a long period before
the new system will have as few bugs as the old one
(and the stuff they run are often very important).
But the exact same reason that keep the systems around
keep the work done to a bare minimum. If a new feature
can be implemented in some middleware in front of the
old system, then that that option is typical picked to
avoid the risk of impacting the core system. Sometimes
new features has to go into the core and that require
the org to keep a minimal staff of COBOL developers.

> There may be a lot of garbage written in Java, PHP and
> even Python, but OSes and, yes, compilers are still being
> written in C.

OS'es, compilers, databases, VM's for the new languages
etc. are still mostly C/C++.

And the competition in this area is not Java, PHP, Python
and similar but languages like Go and Rust.

Arne