Konrad, Okay, I was looking in the wrong place. Which leads me to suggest the following:
Create a local fork of Clojure, make a new branch, and hack on the compiler. Run the experiment, see what happens :) Sean On Feb 24, 9:15 am, Konrad Hinsen <konrad.hin...@fastmail.net> wrote: > On 24.02.2010, at 14:43, Sean Devlin wrote: > > > How does the reader know the difference between .hashCode > > and .reallyObsureMethod? It would need to keep a whitelist of > > everything in object, and know that these methods can be called > > directly. Maybe the reader should be upgraded to handle this? > > It's not the reader who deals with method lookup, it's the compiler. It > doesn't need any whitelist for that. If the compiler knows the (Java) type of > a data item, it can look up the available methods statically, i.e. at compile > time, and generate a direct method invocation is the lookup succeeds. > > This works in fact very well, it's the basis of using hints for avoiding > reflection. What I am surprised about is that in the absence of any hints, > the compiler assumes to know nothing about a value's type, whereas I believe > it could safely assume that it is a subclass of java.lang.Object, considering > that only primitive types are not subclasses of java.lang.Object, but > function arguments cannot be primitive types. But I am not enough of a JVM > expert to be sure. > > Konrad. -- 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
