On Sun, Aug 8, 2010 at 10:50 PM, Meikel Brandmeyer <m...@kotka.de> wrote: > Hi, > > On Aug 8, 6:19 pm, gary ng <garyng2...@gmail.com> wrote: > >> I wrote a similar version in F# which does have the advantage of >> handling infinite input or a very long partition in the sense that I >> can still consume the first element immediately and can skip to the >> second, third group ... In a sense, the result is a lazy list of lazy >> list and if the consumer doesn't hold on to any of them, the memory >> usage is constant. > > I don't think it is possible to do this. At least not in a functional > way. > > Assume your first item of the resulting sequence is a lazy sequence > representing the first group. To define the rest of the sequence you > have to know the end of the first group in the input sequence. However > this can only be found by traversing the input sequence. So this must > again be some lazy sequence like a drop-while or so. But no matter how > it is directly implemented, it must retain the head of the input > sequence for traversal. So, if I traverse the first group, its items > cannot be GC'd because of the rest definition of the sequence. But the > same holds true if I skip to the second group without realising the > first one: to find the actual second group I have to realise the rest > of the sequence, which will also realise the items of the first group. > > So there is no point in not realising the groups: its done anyway, no > matter what. You just introduce unnecessary laziness overhead. > Yes and No.
I have experiemented it a bit back when I was doing it and in certain cases, there are some advantages(mostly in memory usage) even though the front of the list still needs to be realized. Say I want the 'head of each group'. I think the Haskell implementation has this characteristic too. -- You received this message because you are subscribed to the Google Groups "Clojure" group. To post to this group, send email to clojure@googlegroups.com Note that posts from new members are moderated - please be patient with your first post. To unsubscribe from this group, send email to clojure+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/clojure?hl=en
