Fri May 4 18:38:18 EDT 2012


Been a while.  It's starting to itch again.  Looking at those USB
connectors in my rack here..  Maybe it's time for another pass.  Get
the USB driver going and make a synth controller.  Why in that order?
The USB would seriously simplify interface issues, and is long
overdue.  I wonder what Staapl could have been where it not for the
failure to have a working USB interface.  However, the reasons for
that impasse are quite deep, and seeded in doubt about "the right

I still like the idea of an untyped macro language, but currently it
lacks something that is completely trivial in C: hierarchical data
structure namespaces.

There is still the idea of "protocol oriented programming" or "minimal
complexity stream parser approach" whatever name it should bear, but
unfortunately that doesn't work so well with existing protocols like
USB which are based on random access to flat memory buffers, an
approach that is quite biased to using C.

So, to summarize:

- formalize "minimal complexity stream parser protocol design"
- implement hierarchical namespaces
- get that damn USB driver to work

Reading the previous post, it seems that at least the idea of
protocol-oriented programming is a bit stable.  Maybe time to figure
out if it actually makes sense practically.  Theoretically at least I
see a whole bunch of complexity disappear if protocols are designed
better, or even automatically.

And the solution for adding namespaces had crossed my mind too:
there's lambda to introduce local names.  Much more is not needed,
nesting those will do just fine.