|
Fortunately there is a direct and effective process
for converging on the selection of a concept when developing a new product.
Perhaps it can help the team settle the IRML / P issue.
I learned it as "Pugh concept selection" but it is
called "Controlled Convergence" in this reference:
(A more complete description is in Stewart Pugh's
book:
Those who have read both the IRML spec and the P
spec have completed step 1 of the process.
I believe step 2 is a difficult one. Basically the
team has to agree on a list of requirements or specifications for the solution.
This actually seems to be where the debate has centered. Here is a starting
point:
1) Easy to write (by programmers or non
programmers? Using a simple editor, a structured editor, or a GUI? seems like
this has to be decided)
2) Easy to read (by whom, similar set of
issues)
3) Easy to edit and update (by whom, similar set of
issues, do we need hot slide in?)
4) Easy and clear _expression_ of rules logic
including . . . (fill in the blank - I think if / then / else does a lot. what
more is needed? Perhaps Alex's reference set of rules can help
here)
5) easy to write parser / interpreter (this begs
the question of the run time representation - text, byte code, object code, or
else?)
6) easy / fast to run (how fast is
needed?)
7) Compact storage on the processor
8) etc.
9) etc.
10) etc.
I propose the working group continue to debate the
list of requirements, (perhaps using the above list as a starting point.)
When the list of requirements converges, I would be happy to lead the team
through the remaining steps of the convergence process.
Is this sensible and helpful?
Lee Beaumont
|