> On Jun 17, 2026, at 1:21 PM, David Wade via cctalk <[email protected]>
> wrote:
>
>
>
> On 17/06/2026 15:56, Bill Gunshannon via cctalk wrote:
>>
>>
>> On 6/17/2026 10:51 AM, Johan Helsingius via cctalk wrote:
>>> On 17/06/2026 4:49 pm, Bill Gunshannon via cctalk wrote:
>>>> I came on the scene of a Honeywell to Univac-1100 migration 46 years
>>>> ago. Mostly COBOL a little Fortran and a tiny bit of other silly
>>>> stuff. Not only was there no thought of converting the COBOL to Ada
>>>> at the time but the last time I checked they are still running some
>>>> of the COBOL I wrote when I was there.
>>>
>>> Sure, but would anyone starting a new software project choose
>>> COBOL as the implementation language?
>>>
>>
>> Other than the dearth of programmers, why not? What's wrong with the
>> language when the project is in its domain?
>>
> Nothing, but we will disagree on defining its domain which I believe must
> include hardware and software.
> So if I am extending something written in COBOL on an IBM Mainframe or
> Mid-range then yes, that Is the domain of COBOL.
> For something thats going to run in the cloud with a web front end unlikely.
No, the domain for COBOL is an application domain. It can be run on any
machine that has a compiler for it. It's true that some machines (IBM/360,
VAX) have instructions specifically designed to do the decimal arithmetic that
COBOL likes as single CISC operations, but the lack of specific instructions
like that has never been a compiler limitation. No more than the lack of a
hardware stack keeps you from writing recursive applications, or prevents
compilers from compiling such programs.
paul
