On Friday, 27 November 2015 at 12:16:34 UTC, Jonathan M Davis
wrote:
Well, if you're dealing with a pseudo-random generator with a
specific seed, I'm not sure that it's okay, though obviously
for a fully random number generator, it wouldn't matter.
I think the distinction here is artificial. If `range` is a
random number generator, then `take` has no right to expect its
output to be deterministic or consistent in any way.
When we come to input ranges in general, that's surely a factor
we need to take into account. Can the InputRange actually be
relied on as a deterministic iteration, but one where we can't
save our state? Or should it be treated as simply a source of
data, about which one can make no assumptions regarding its
consistency?
I think answering that question -- and maybe distinguishing the
two cases if possible -- feeds into how to address the subsequent
problems you describe.
Possibly, but because almost everything in Haskell is both
functional and lazy, you don't really get the problem of
popFront being called after the call to take.
I'm not sure that's really true. I'm reasonably sure that I can
(given some source of IO) do something like (pseudo-code because
my Haskell is rusty),
do
lazyData =
something_that_lazily_reads_5_entries_from_the_IO_entity
read_some_data_from_IO
read_some_data_from_IO
do_something_with_lazyData
If the source of IO is the standard random number generator of
Haskell, for example, it'll behave like the input range/popFront
example.
