Talk:Sieve of Eratosthenes: Difference between revisions
→Recent edits to Haskell and my reversal of them
GordonBGood (talk | contribs) (→Recent edits to Haskell and my reversal of them: making functions instead of simple bindings...) |
|||
Line 272:
:::: I see your point on using the combinator to avoid holding the lists in memory '''for the actual sequence of calculations''', however making each a function as in "primes :: () -> [Int]" rather than "primes :: [Int]" does two things: 1) for noobie people testing timing and such, it means that on every call to the '''function''' the whole calculation sequence starts from the beginning rather than using that portion of the memoized stream already computed, and 2) having the outer binding means that the result stream can not be garbage collected as the stream is consumed by say "head $ drop 1000000 primes" where I am suggesting "head $ drop 1000000 $ primes ()" will not hold onto the results stream and will have a very low memory residency; one does have to be careful when making recursive references (not using "fix" or combinators) as a reference to a function does generate a whole new stream - when that is not desired, internal simple bindings must be used to avoid the generation of new streams when they are not desired. --[[User:GordonBGood|GordonBGood]] ([[User talk:GordonBGood|talk]]) 00:10, 12 April 2015 (UTC)
::::: but those versions are inefficient anyway. Plus, the proper way to test is to run a compiled, standalone executable, where even <code>primes :: [Int]</code> is garbage collected and runs in near constant memory. But here I thought the aim is to showcase a language not even to (practicing) newbies (there are plenty more other resources for those), but to people even totally unfamiliar with it, just to glance over (but the snippets are still runnable of course). So for me, clarity was the utmost goal. [[User:WillNess|WillNess]] ([[User talk:WillNess|talk]]) 12:06, 12 April 2015 (UTC)
| |||