I certainly like that better than the binding solution, especially if hotspot can potentially inline the functions. It still tricky to do that in a macro, especially when the code to be evaled depends on the current lexical scope, but I can maybe work around that. Thanks.
On 29 August 2013 04:50, Jason Wolfe <ja...@w01fe.com> wrote: > Sounds interesting, I'd love to hear more about the use case. > > We ran into this issue as well when implementing the new positional > compilation for Graph. We filed a ticket [1], and worked around it with a > funny trick: > > Suppose you want to eval a complicated thing, and the crux of the issue is > that this fails: > > (let [y 1 f (fn [x] (+ x y))] (eval `(~f 1))) > > A solution is to rework what you eval, so that it has a 'hole' for the > function and you slot the function into this hole afterwards (which Hotspot > seems to be able to optimize just fine): > > (let [y 1 f (fn [x] (+ x y))] ((eval `(fn [g#] (g# 1))) f)) > > Happy to elaborate if you like. > > Cheers, > Jason > > [1] http://dev.clojure.org/jira/browse/CLJ-1206 > > > On Wednesday, August 28, 2013 10:59:19 AM UTC-7, Jamie Brandon wrote: >> >> That sort of works for my use case. >> >> What I want to do is define a grammar... >> >> (def num-graph >> (graph >> num ~(or ~succ ~zero) >> succ (succ ^x ~num) >> zero zero)) >> >> ... attach actions .... >> >> (def num-out >> (output-in num-graph >> 'zero (fnk [] 0) >> 'succ (fnk [x] (inc x)))) >> >> ... and compile the result ... >> >> strucjure.regression.tests> (graph->view 'num num-out) >> (clojure.core/letfn >> [(num ...) >> (succ ...) >> (zero ...)] >> (clojure.core/fn >> [input__2288__auto__] >> (num input__2288__auto__))) >> >> This works fine as long as the fnks dont close over anything, but >> that's very limiting. >> >> If I eval that code and *then* wrap it in (binding ...) I can have >> closures but that now means that I can't do call-site compilation. >> I'll have to poke around a bit more... >> >> On 28 August 2013 18:32, Aaron Cohen <aa...@assonance.org> wrote: >> > I'm not sure if you'll consider this hacky or not. >> > >> > (def ^:dynamic *fn-helper*) >> > >> > (defn eval-at-one [f] (binding [*fn-helper* f] (eval '(*fn-helper* 1)))) >> > >> > >> > On Wed, Aug 28, 2013 at 1:28 PM, Jamie Brandon >> > <ja...@scattered-thoughts.net> wrote: >> >> >> >> You will also run into problems if your functions close over any data: >> >> >> >> user> (let [f (fn [] 1)] (eval `(~f))) >> >> 1 >> >> nil >> >> user> (let [x 1 f (fn [] x)] (eval `(~f))) >> >> IllegalArgumentException No matching ctor found for class >> >> user$eval38616$f__38617 clojure.lang.Reflector.invokeConstructor >> >> (Reflector.java:163) >> >> >> >> I'm struggling with this as well, I'm definitely interested in finding >> >> a non-hacky way to pass closures through eval without leaking memory. >> >> >> >> Perhaps something like: >> >> >> >> (let [x 1 >> >> f (fn [] x) >> >> f-sym (gensym)] >> >> (intern *ns* gensym (weak-ref f)) >> >> (with-meta (eval `(fn [] (do-stuff-with @~f))) {::strong-ref f})) >> >> >> >> So that way f will be freed up when the resulting fn is gced. >> >> >> >> On 28 August 2013 17:59, Ben Wolfson <wol...@gmail.com> wrote: >> >> > On Wed, Aug 28, 2013 at 9:27 AM, Jamie Brandon >> >> > <ja...@scattered-thoughts.net> wrote: >> >> >> >> >> >> >> >> >> If you aren't worried about leaking memory, a possible workaround is >> >> >> to intern the fn in some namespace and then put the symbol in the >> >> >> macro output. >> >> > >> >> > >> >> > Hrm, I hope it doesn't come to that---as long as I'm creating the >> >> > functions >> >> > with their metadata, I can make things work, because I can add extra >> >> > information to the with-meta call or do extra things when making the >> >> > call. >> >> > What I'd like, though, is for it to be possible to copy the metadata >> >> > over to >> >> > a new function by client code, just using the regular (with-meta (fn >> >> > [blah] >> >> > blah) (meta old-fn)) mechanism. But then the sneaky symbol, or >> >> > whatever >> >> > extra thing, in the metadata, that's used in the form to be eval-ed >> >> > is >> >> > pointing to the original function, not the one passed in. Maybe just >> >> > providing a utility for fixing up the metadata is the right move, >> >> > though >> >> > it's kidn of dissatisfying. >> >> > >> >> > >> >> > -- >> >> > Ben Wolfson >> >> > "Human kind has used its intelligence to vary the flavour of drinks, >> >> > which >> >> > may be sweet, aromatic, fermented or spirit-based. ... Family and >> >> > social >> >> > life also offer numerous other occasions to consume drinks for >> >> > pleasure." >> >> > [Larousse, "Drink" entry] >> >> > >> >> > -- >> >> > -- >> >> > 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/groups/opt_out. >> >> >> >> -- >> >> -- >> >> 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/groups/opt_out. >> > >> > >> > -- >> > -- >> > 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/groups/opt_out. > > -- > -- > 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/groups/opt_out. -- -- 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/groups/opt_out.
