Re: The joy of FORTRAN

The Natural Philosopher <tnp@invalid.invalid> wrote:

On 01/10/2024 19:44, rbowman wrote:

On Tue, 1 Oct 2024 00:56:56 -0400, 186282@ud0s4.net wrote:
 

Technically, I agree. However, as mentioned, PRACTICAL issues
intervene. Time/MONEY/reliability are also very important. These
ain't the filthy-rich 60s anymore so if it WORKS you DON'T mess with
it.

 
There comes a point in many projects when you realize that the effort
helped define the criteria and while you have a better understanding of
the problem you really should scrap the code and start over. Time, money,
and ego involvement ensure that seldom happens.

 
Ive done that several times.
You sketch out something, realise that was partly wrong, or has got
messy, or could be done more simply if chunked out to subroutines, and
start again using the bits of code that worked.
 
The cost benefit is against the top down "we will write the whole
detailed spec before we write a line of code" idea  that in my
experience is actually worse.
 
In practice I work both ends to the middle. Write what is obvious in
spec or code first, and then see what problems remain unsolved.....
 

I’ve done it a few times. I thoughhtI had a design that would work, but
after a bit of coding problems showed up, so I scrapped it and started
over. I think better in code than in the abstract, so this was my
equivalent of doodling on a yellow pad. Also, it was almost never a total
loss - most of the functional routines were reusable, and only the logic
needed to be redone.

--
Pete