Wed Jun 3 23:14:12 CEST 2009
I think I've been doing this for a long time now. For me it's
always been network system administration, improving debugging tools,
starting new projects from crazy ideas and generally structured
talking out of my ass to see what it brings..
Actually, apart from te net admin it's maybe all true
Lately I've been wanting to study more PL theory, but I still find it
more interesting to re-invent the weel following subtasks in Staapl.
The Forth bootstrap has been quite a funny example here.. It's an
interesting problem. Not only for nostalgic reasons (yeah those 8 bit
thingies) but I like building non-trivial things on simple systems -
to think, where mere brute force programming pattern duplication would
be much more economic with time.. But that, is excepting defeat.
Anyway.. What am I saying.. This is not a blog.. It's about the code,
The real problem I'm avoiding at this moment is partial evaluation and
the functional concatenative language. I also promised myself to
write the USB driver first, and write the documentation. But instead
I'm bootstrapping a standard Forth on top of a system that is
distinctively non-standard for very good reasons. Why? Because it's
fancy and shows off the flexibility of the system.
The PE is difficult.. Maybe mostly because the problem is not
well-defined. It's difficult to define "optimal". People do seem to
have tried (and maybe succeeded?). Hence I didn't read too much yet
The USB driver is difficult for a different reason: it's extremely
tedious and error-prone to write in a direct style. I'm trying to do
it a bit more abstract so I can learn a thing or two for the next
complicated driver I need to write.. I'm thinking about specializing
in driver writing, and write some tools for that. It's a difficult
problem worthy of some attention. It's also particularly unglamorous,
so might make me some money in the process.
Writing documentation is difficult. Writing clearly tout court is
difficult. I'm starting to gain more and more respect for good
teachers. And for good manuals. The PLT Scheme manual is a good
So, what for tomorrow? Maybe finish the bootstrapping.. :)