[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: The problem(s) with threads



> Hans Boehm wrote a paper about that:
> Âhttp://portal.acm.org/citation.cfm?doid=1065010.1065042


I haven't seen anything other than his abstract, but this seems to be
exactly the challenge to the received wisdom on libraries I had in
mind.

Your remarks on type system design and error messages makes me wonder:
does this indicate that our intellectual effort to design better type
systems needs to bifurcate into effort in type systems and separate
effort in usage rules.  The usage rules are about how program
constructs can fail certain criteria that yield clear error messages.
I'm thinking Occam compiler of old, here.  It's different from
designing types, superficially at least.

A hypothetical language that allowed usage rule declarations would
address the extensibility problems faced by the Kroc team et al - many
good concurrency primitives were invented, but each required new
language syntax and new compiler effort.  This is far from ideal.  One
of Scala's objectives is being able to use its flexible syntax so that
libraries appear to be extending the language, thus getting round the
language's scalability problem.  But as you said for Haskell, if you
get too smart, you get indecipherable error messages.  This is no way
forward; there ought to be a better way.

My 2p  :)
Rick Beton