Dataflow and Comonads was Re: [Haskell-cafe] Monads in Scala, ...

[Bulat Ziganshin]

Hello Bill,

Sunday, November 27, 2005, 1:25:59 AM, you wrote:

BW> The one downside I found to using dataflow was that most software people
BW> seem to be uncomfortable with the lack of identifiable processes doing
BW> significant bits of work.  I guess if they they're not floundering
BW> around in mutual exclusion, semaphores, deadlock detection and all the
BW> other manifestations of unmanaged complexity, they don't feel they've
BW> *accomplished* anything (BTW I grew up on Dijkstra, Hoare and Hanson, so
BW> I can get away with saying this :-).  Interestingly enough, and perhaps
BW> obvious in retrospect, I often found hardware designers to be very
BW> comfortable with dataflow computations.

dataflow computers was known at least from 60's as possible
alternative to Neumann architecture with its bottleneck of only one
operation executed each time. they are very natural for chip designers
because real internal processor structure is dataflow, and for
external users (assembler programmers) Neumann architecture is
emulated



-- 
Best regards,
 Bulat                            mailto:bulatz at HotPOP.com