Re: The joy of FORTRAN

On 10/1/24 10:12 PM, rbowman wrote:

On Tue, 1 Oct 2024 21:28:14 +0200, D wrote:
 

I think there's some old programmer saying from the beginning of time,
that says something like the third time I completely rewrite the program
it gets done right.

 Over the years new buzzwords like agile have been tacked on but my process
always started with the assumption that the client didn't know what they
really wanted. Throw together a quick and dirty prototype with enough
functionality that they can play with it. Take notes on the complaints,
modify the prototype. Rinse and repeat until they're mostly happy.
 Scrap the prototype, possibly retaining some code, and build the product.
 There is a related process I call 'reconnaisance by fire'. That's when
you're handed an API that you're pretty sure is incomplete, inconsistent,
and missing vital information. Build the interface, test it, find out
where it fails, and try something else. If you're lucky the other party
will be helpful although I have had instances where they were hostile.
 The most important part is not developing an ulcer while dealing with
human nature. Easily perturbed people probably should try another career
path.

   Clients want "something" - but they're often not sure
   what that will look and feel like. They will ALWAYS
   have complaints, ALWAYS want more and more 'features',
   some of which will be incompatible with each other,
   added and may run off to somebody else expecting magic.
   Been there, seen that.
   Alas this means your "final" code will wind up as a MESS
   that's hard to read/debug. You can't keep throwing out
   everything, money issues prevent that, the world wasn't
   built for programmers - mostly for bean-counters these days.
   The "third time" story IS mostly correct BTW :-)