On Mon, Aug 31, 2009 at 10:55 AM, Krukow<karl.kru...@gmail.com> wrote:
>
> I have two minor minor suggestions for Clojure changes.
>
> 1) Consider this function:
> user> (set! *warn-on-reflection* true)
> true
> user> (defn reader-from-classpath [s]
> (-> (.getResourceAsStream java.lang.String s)
> (java.io.InputStreamReader.)
> (java.io.BufferedReader.)))
> Reflection warning, NO_SOURCE_PATH:2 - call to getResourceAsStream
> can't be resolved.
> #'user/reader-from-classpath
>
> In general, I think every call of form (.instanceMember Classname
> args*) will generate such a warning since it expands to, e.g.,
>
> user> (macroexpand ' (.getResourceAsStream java.lang.String s))
> (. (clojure.core/identity java.lang.String) getResourceAsStream s)
> user>
>
> And identity doesn't have any type information. A simple fix would be.
> user> (defn #^Class class-identity [#^Class c] c)
>
> #'user/class-identity
>
> and then expanding (.instanceMember Classname args*) to
>
> (. (class-identity Classname) instanceMember args*)
>
> -------
>
> 2) I can do:
> user> (into {} '([:k :v]))
> {:k :v}
>
> This works for two-element vectors. However, I cannot do the same for
> two-element arrays:
>
> user> (def str_array (.split "k=v" "="))
> #'user/str_array
> user> (into {} (list str_array))
> ; Evaluation aborted.
>
> It would be a simple addition to clojure.lang.ATransientMap:
> import java.lang.reflect.Array;//change
> ...
> public ITransientMap conj(Object o) {
> ensureEditable();
> if(o instanceof Map.Entry)
> {
> Map.Entry e = (Map.Entry) o;
>
> return assoc(e.getKey(), e.getValue());
> }
> else if(o instanceof IPersistentVector)
> {
> IPersistentVector v = (IPersistentVector) o;
> if(v.count() != 2)
> throw new IllegalArgumentException("Vector arg
> to map conj must be
> a pair");
> return assoc(v.nth(0), v.nth(1));
> }//begin change
> else if(o != null && o.getClass().isArray())
> {
> if(Array.getLength(o) != 2)
> throw new IllegalArgumentException("Array arg
> to map conj must
> have exactly two elements");
> return assoc(Array.get(o,0), Array.get(o,1));
> }//end change
>
> ITransientMap ret = this;
> for(ISeq es = RT.seq(o); es != null; es = es.next())
> {
> Map.Entry e = (Map.Entry) es.first();
> ret = ret.assoc(e.getKey(), e.getValue());
> }
> return ret;
> }
>
> After this change I have:
> user> (def str_array (.split "k=v" "="))
> #'user/str_array
> user> (into {} (list str_array))
> {"k" "v"}
> user>
>
> The use-case arose from wanting to create a map from a properties file
> using split ("=") on each line of the properties file (and not wanting
> to copy the array into []).
>
> N.B. the cost is an additional null-check and o.getClass().isArray().
>
> What do you guys think?
#1 is fine idea. I've implemented the hinting in the compiler where
that expansion takes place. (commit e45046da8f)
As for #2, I'm less sure I want to encourage people to interoperate
with the map library using arrays.
Rich
--~--~---------~--~----~------------~-------~--~----~
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
-~----------~----~----~----~------~----~------~--~---
