[REBOL] Re: More on expressions
From: sunandadh:aol at: 8-Feb-2003 15:50
Hi Joel,
Thanks for a comprehensive and interesting discussion as usual. You are right
that I attempted to move the issue from algorithm design to application
design. I guess that's my bias.
I enjoy the fact that we can come up with so may algorithms for doing the
same thing. We need lots of different algorithms, each with its own
datasheet
of strengths and weaknesses. That way, application designers can
pick the one they need.
Different application designers will pick different algorithms depending on
their needs. The right solution for a real-time controller (needing a
determinate response time) may be different from another embedded-chip
application (needing smallest footprint) and so on.
Me, I've been warped by three decades of commercial application design. Means
I normally instinctively go for the solution with the fewest built-in
assumptions -- the most "general" usual, but seldom the most elegant.
I know -- believe me I do -- that within 24 hours of implementing a median
algorithm that handled numerics, I'd be on the wrong end of this phone call:
It's broken! It's not displaying the median.....Yes, I know we said they'd
be numbers -- but not always....On a Tuesday afternoon of course the three
things could be city names....No of course we don't want the middle one
alphabetically in that case -- why on earth would anyone want *that*!?....We
need the median *population* -- unless the user logged on is a member of the
ethics committee then they'd probably want the city with the median distance
from the North Pole -- but check with them first, that's only a guess.....Why
should I know if there are any other
special cases" like that? -- you wrote
the code: it's your job to know what median means.....And we need it fixed
within 20 minutes."
Sunanda.
