Le samedi 11 octobre 2014, Jan-Paul Bultmann <janpaulbultm...@googlemail.com> a écrit :
> Yes you are right, currently a lot of stuff is seq only. > I think that reducers provide a lot of eager operations, but not > everything. > > So yeah, transducers are exactly what you are looking for. > > If you call `sequence` with the transducer you get laziness, if you call > `into` with a collection and a transducer, you get eagerness :). > No decision about the evaluation strategy has to be made up until that > point, and afaik transducers implement most classical seq constructs. > Yep. There is a page dedicated to transducers and it lists (near the middle) the core functions that have a new arity for returning transducers: map <http://clojure.github.io/clojure/branch-master/clojure.core-api.html#clojure.core/map> cat <http://clojure.github.io/clojure/branch-master/clojure.core-api.html#clojure.core/cat> mapcat <http://clojure.github.io/clojure/branch-master/clojure.core-api.html#clojure.core/mapcat> filter <http://clojure.github.io/clojure/branch-master/clojure.core-api.html#clojure.core/filter> remove <http://clojure.github.io/clojure/branch-master/clojure.core-api.html#clojure.core/remove> take <http://clojure.github.io/clojure/branch-master/clojure.core-api.html#clojure.core/take> take-while <http://clojure.github.io/clojure/branch-master/clojure.core-api.html#clojure.core/take-while> take-nth <http://clojure.github.io/clojure/branch-master/clojure.core-api.html#clojure.core/take-nth> drop <http://clojure.github.io/clojure/branch-master/clojure.core-api.html#clojure.core/drop> drop-while <http://clojure.github.io/clojure/branch-master/clojure.core-api.html#clojure.core/drop-while> replace <http://clojure.github.io/clojure/branch-master/clojure.core-api.html#clojure.core/replace> partition-by <http://clojure.github.io/clojure/branch-master/clojure.core-api.html#clojure.core/partition-by> partition-all <http://clojure.github.io/clojure/branch-master/clojure.core-api.html#clojure.core/partition-all> keep <http://clojure.github.io/clojure/branch-master/clojure.core-api.html#clojure.core/keep> keep-indexed <http://clojure.github.io/clojure/branch-master/clojure.core-api.html#clojure.core/keep-indexed> dedupe <http://clojure.github.io/clojure/branch-master/clojure.core-api.html#clojure.core/dedupe> random-sample <http://clojure.github.io/clojure/branch-master/clojure.core-api.html#clojure.core/random-sample> http://clojure.org/transducers HTH > > On 11 Oct 2014, at 21:09, Mars0i <marsh...@logical.net > <javascript:_e(%7B%7D,'cvml','marsh...@logical.net');>> wrote: > > I don't think I have any misconceptions. I probably haven't expressed > myself clearly. Clojure has many very useful functions that return lazy > sequences. Analogous functions in other languages perform what are exactly > the same operations, except for the fact that what's returned is not a lazy > sequence, but instead is a different kind of sequential structure. For > example, Common Lisp'smapcar is very much like Clojure's map, but mapcar > returns > a (non-lazy) singly-linked list. As you point out, there are also > functions in Clojure that return sequential structures that are not lazy. > Moreover, it's easy to convert a lazy sequence into a non-lazy sequence > through use of, for example, doall or vec. > > However, writing code in Clojure that *only* used non-lazy sequences > would be unnatural and verbose, as Lee points out. No one would bother > doing that. Part of what's great about Clojure is the convenience of its > set of sequence operations. If you're not using those, you're not getting > full benefit from Clojure. So as a practical matter, one must deal with > laziness in Clojure. > > However, almost the same language could be constructed without any > laziness, just by replacing lazy functions such as map with non-lazy > analogues. In that case, you could do all of the same things in those > situations that don't require laziness. (You couldn't use lazy infinite > sequences, and you'd have to worry about keeping large data structures in > memory, etc. But in many situations, those features aren't needed.) > > That's what I mean by "turning off laziness": Replacing every function > that returns a lazy sequence with its closest non-lazy analogue. > > On Saturday, October 11, 2014 12:01:34 PM UTC-5, Jan-Paul Bultmann wrote: >> >> I feel that there are some misconceptions here, >> Clojure is not a language that has lazy evaluation, >> but is instead completely eager. >> Macros are the closest thing that come to laziness, but they are still >> different. >> However, Clojure has a data structure, that also serves as an abstraction >> layer, >> that implements laziness, called a sequence or seq. >> >> Therefor “turning of laziness” makes no real sense. You could say >> “disable the laziness of seqs”, >> but that doesn’t really make sense either. It’s like disabling the FIFO >> nature of a queue or a channel. >> If you don’t want the properties of a seq, then use another data >> structure, like a vector. >> >> >> On 11 Oct 2014, at 18:14, Mars0i <mars...@logical.net> wrote: >> >> Thanks Jan-Paul. That's helpful. I wonder whether, if all lazy >> functions were rewritten in terms of transducers, it would then be easy to >> turn laziness on and off. (I am not suggesting that this should be done. >> No doubt it would entail a lot of work, and performance tradeoffs. I'm >> quite happy with Clojure as it is, despite anything negative I might say >> about impacts of laziness in some circumstances.) >> >> On Saturday, October 11, 2014 5:00:36 AM UTC-5, Jan-Paul Bultmann wrote: >>> >>> Transducers build up transformation steps, with new generic operations >>> like single arg `map` and `filter`, and then apply these transformations >>> to a collection using an evaluation function. >>> This function decides if operations are done lazily, eagerly, or on a >>> channel. >>> >>> Note that while this makes the place where laziness occurs more obvious, >>> it is not really new. >>> You can currently infer as well that `map` `filter` `concat` e.t.c are >>> lazy. While `mapv` or `filterv` are not. >>> >>> >>> On 11 Oct 2014, at 00:28, Mars0i <mars...@logical.net> wrote: >>> >>> >>> >>> On Friday, October 10, 2014 5:20:30 PM UTC-5, Mars0i wrote: >>>> >>>> Maybe an ideal world would be one in which there was a global setting >>>> to turn laziness on and off. When you want it, have it, and know your >>>> risks. After looking at the source for some of the lazy functions, I've >>>> come to suspect that such a feature would be completely nontrivial. >>>> >>> >>> Oh, wait, Rich Hickey's blog post about transducers >>> <http://blog.cognitect.com/blog/2014/8/6/transducers-are-coming> says: >>> >>> >>> >>> >>> >>> *But transducers can also be used for: a la carte laziness ... >>> collection/iteration/laziness-free transforming reductions * >>> >>> Not certain what this means. Haven't fully grokked reducers. >>> >>> -- >>> You received this message because you are subscribed to the Google >>> Groups "Clojure" group. >>> To post to this group, send email to clo...@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+u...@googlegroups.com >>> For more options, visit this group at >>> http://groups.google.com/group/clojure?hl=en >>> --- >>> You received this message because you are subscribed to the Google >>> Groups "Clojure" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email toclojure+u...@googlegroups.com. >>> For more options, visit https://groups.google.com/d/optout. >>> >>> >>> >> -- >> You received this message because you are subscribed to the Google >> Groups "Clojure" group. >> To post to this group, send email to clo...@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+u...@googlegroups.com >> For more options, visit this group at >> http://groups.google.com/group/clojure?hl=en >> --- >> You received this message because you are subscribed to the Google Groups >> "Clojure" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to clojure+u...@googlegroups.com. >> For more options, visit https://groups.google.com/d/optout. >> >> >> > -- > 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 > <javascript:_e(%7B%7D,'cvml','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 > <javascript:_e(%7B%7D,'cvml','clojure%2bunsubscr...@googlegroups.com');> > For more options, visit this group at > http://groups.google.com/group/clojure?hl=en > --- > You received this message because you are subscribed to the Google Groups > "Clojure" group. > To unsubscribe from this group and stop receiving emails from it, send an > email toclojure+unsubscr...@googlegroups.com > <javascript:_e(%7B%7D,'cvml','clojure%2bunsubscr...@googlegroups.com');>. > For more options, visit https://groups.google.com/d/optout. > > > -- > 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 > <javascript:_e(%7B%7D,'cvml','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 > <javascript:_e(%7B%7D,'cvml','clojure%2bunsubscr...@googlegroups.com');> > For more options, visit this group at > http://groups.google.com/group/clojure?hl=en > --- > You received this message because you are subscribed to the Google Groups > "Clojure" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to clojure+unsubscr...@googlegroups.com > <javascript:_e(%7B%7D,'cvml','clojure%2bunsubscr...@googlegroups.com');>. > For more options, visit https://groups.google.com/d/optout. > -- Laurent Petit -- 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 --- You received this message because you are subscribed to the Google Groups "Clojure" group. To unsubscribe from this group and stop receiving emails from it, send an email to clojure+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
