[<<][staapl][>>][..]Sun Nov 4 15:53:35 CET 2007
embedded programming in 2007
the question i really like to answer: without too much bias (the the
tool i wrote) what is the point of writing static, early bound code in
2007, even if we're talking about microcontrollers.
* is there really a 'complexity barrier' below which one HAS TO move
to quasi manual compilation and allocation?
* will this barrier remain in existence, or will better tools make
a more high-level approach possible?
EDIT: some things i was thinking about yesterday:
* leaky abstractions are hard to work with. starting from assembly
and "thinking up", using purrr to help you write the application
is the right approach. starting from some high-level understanding
of the language and having to learn all its limitations doesn't
really work. the problem seems to be the manual resource
management: time, space, and synchronization between global
variables, and hardware devices.
* it seems i loose most of my time in low-level configuration issues
which give little feedback on error, and dealing with situations
that are hard to debug due to dependence on external events. low
level design really is a debugging problem: setting up experiments
to try to isolate errors. hence the use of loads of specialized
(hardware) tools used in professional environments.