[<<][meta][>>][..]
Wed Sep 23 10:41:20 CEST 2009

code generation from UML

Since UML[1] seems to be the paradigm around which a whole lot of
model-based approaches are built, it's time to give it a good look.

From the more approachable FP camp, I find this LtU tread[2].  A
characterization of modeling languages is to give a platform to talk
about strategies instead of implementations.  

Anton van Straaten complains about representation-oriented approach in
standard OO use (abstraction violations), and later about the
evolutionary nature of these kinds of tools (the pyramid).

Paul Snively notes that the gap between design and programming narrows
as type systems get better.  Marc Hammann talks about the ``undead
monster'' (OO vs. FP) and calls the OO/FP distinction (object/ADT,
class/module) an implementation detail. ``I think I understand what
problem those people think they are solving: 1) Design is hard.  2)
Many programmers lack good design skills.  3) Therefore, we must
encode good design into tool and languages so they don't need to
develop these skills.'' Allan McInnes points to FAD[3].

[1] http://en.wikipedia.org/wiki/Unified_Modeling_Language
[2] http://lambda-the-ultimate.org/node/1828
[3] http://www.cs.kent.ac.uk/pubs/2001/1152/
[4] http://www.ncgia.ucsb.edu/conf/interop97/program/papers/kuhn.html
[5] http://en.wikipedia.org/wiki/SCADE



[Reply][About]
[<<][meta][>>][..]