the new math: old battle of the lovees was: PDP1 3601
 |
I know. I know. Theoreticians create new proofs or logics using the above tools. They work on stuff that hasn't been done ever before. They help build...
There is something in people's minds that react to functional languages.
I have tried to push collegues in the direction of more functional programming in several programming projects. This is one of the places I see people really struggle with the concepts.
I have written an number of small packages that have small "semi-functional" languages inside. Usually a few C-based state machines, and the actual code in a lightweight functional language "config file".
I made a very tight X.25 implementation this way; and made a little language that dealt with reactions to events as X.25 changed state.
snip -- stuff about Dijkstra replied to in another post more snip -- the list here is of background for CS When I say "proofs" I mean basic...
Similarly, I made a transction router-dispatcher for a trading realtime system that was very similar inside, and the same type of language handling events in a functional language fashion.
These are the two pieces of code I am most proud of; they were both extensively documented, and I put together a library to read up on the theory.
It is stronger than desire. I get twitchy. I have to know. I may even sacrifice a first born to find the...
They were both pretty small; around 2300 LOC for X.25, 800 of which were driver code. Add around 400 LOC of "config" for the protocol itself.
I tried really hard to make a good handover to people that should have had the skillset to pick it up. (CS and math doctorates were plentiful)
But I ended up as the maintainer of both for a good 7 years after I quit that job. The job was in the financial services industry, and all the systems in question are retired now.
The one thing that people really struggled with was the fact that the "imperative" programming model was completely abandoned for a different paradigm. Code flow didn't express the logic anymore, the state machines and functional descriptions did.
Those who loved the systems were the financial whiz kids; they instantly saw the value of the functional descriptions of the protocols; and understood what was going on even if they knew nothing about the actual underlying hardware.
And, indeed, people take very easily to spreadsheets. Perhaps a spreadsheet-like system where functional code snippets can be integrated will be a solution?
We still lack a good, widespread structural concept for the cases where a goto is really what you want.
-- mrr