I *think* it is expected behavior. Take this with a grain of salt, but I *think* that defs are namespace resolved at compile time, and so will not be checking in-ns calls within a function to determine the appropriate namespace, but will instead always use the ns of the function containing the def.
- Mark On Tue, Aug 16, 2011 at 2:54 PM, Richard Rattigan <ratti...@gmail.com> wrote: > Thanks for your replies. Do you think this is a bug, given that the > documentation doesn't seem to concur with this behaviour? > > On Aug 15, 9:54 pm, Mark Rathwell <mark.rathw...@gmail.com> wrote: >> Wow, forget everything I said, this has nothing to do with macro >> expansion. Looks more like inside a function you can only def >> something in the same namespace as the function: >> >> user> (defn ff [] (in-ns 'foo.core2) (def anything5 10)) >> #'user/ff >> user> anything5 >> >> Var user/anything5 is unbound. >> [Thrown class java.lang.IllegalStateException] >> >> user> (ff) >> #'user/anything5 >> foo.core2> (in-ns 'foo.core2) >> #<Namespace foo.core2> >> foo.core2> (def anything6 10) >> #'foo.core2/anything6 >> >> >> >> >> >> >> >> On Mon, Aug 15, 2011 at 9:36 PM, Mark Rathwell <mark.rathw...@gmail.com> >> wrote: >> > You are correct, I must have messed something up in my expansions. >> > From the below, however, it looks like a var is being declared just by >> > having the macro called, but not bound to the value: >> >> > user> (defmacro foo [name & body] `(def ~name ~(identity `(fn [] ~@body)))) >> > #'user/foo >> > user> (foo xx (in-ns 'foo.core) (def anything3 10)) >> > #'user/xx >> > user> anything3 >> >> > Var user/anything3 is unbound. >> > [Thrown class java.lang.IllegalStateException] >> >> > user> anything4 >> >> > Unable to resolve symbol: anything4 in this context >> > [Thrown class java.lang.Exception] >> >> > On Mon, Aug 15, 2011 at 8:56 PM, Alan Malloy <a...@malloys.org> wrote: >> >> I either disagree or don't understand. The deftest macro doesn't touch >> >> your &body arg; it's expanded as-is. For example, (let [x 'foo] `(inc >> >> ~x)) doesn't result in foo getting qualified, and most macros behave >> >> the same way. >> >> >> On Aug 15, 4:36 pm, Mark Rathwell <mark.rathw...@gmail.com> wrote: >> >>> Just to be clear, it is namespace resolved because of syntax quote: >> >> >>> (defmacro deftest >> >>> [name & body] >> >>> (when *load-tests* >> >>> `(def ~(vary-meta name assoc :test `(fn [] ~@body)) >> >>> (fn [] (test-var (var ~name)))))) >> >> >>> On Mon, Aug 15, 2011 at 7:23 PM, Alan Malloy <a...@malloys.org> wrote: >> >>> > Is it? That's neat; I guess I've never thought about how the compiler >> >>> > treats def. Thanks for the explanation. >> >> >>> > On Aug 15, 3:03 pm, Mark Rathwell <mark.rathw...@gmail.com> wrote: >> >>> >> deftest is a macro. Macros are expanded at compile time. So, in this >> >>> >> case, at compile time, a function called namespace2 is def'd with meta >> >>> >> data :test set to the body of your deftest. >> >> >>> >> All of that body is namespace resolved in macro expansion, before >> >>> >> in-ns is ever executed (which happens when you actually call the >> >>> >> namespace2 function created by the macro). Put another way, (def >> >>> >> anything 10) is namespace resolved to (def >> >>> >> learn.clojure.test.core/anything 10) at macro expansion time (compile >> >>> >> time), before the test function is ever called, and thereby before >> >>> >> in-ns is ever executed. >> >> >>> >> Hope this helps. >> >>> >> On Mon, Aug 15, 2011 at 5:07 PM, Richard Rattigan >> >>> >> <ratti...@gmail.com> wrote: >> >> >>> >> > I'm finding that namespaces don't seem to behave as I expect >> >>> >> > intuitively, or according to the reference. It's quite possible I'm >> >>> >> > in >> >>> >> > the wrong here though, as I'm just kicking clojure's tires at this >> >>> >> > point. >> >> >>> >> > Here is the relevant doc: >> >> >>> >> >http://clojure.org/special_forms >> >>> >> > (def symbol init?) >> >>> >> > Creates and interns or locates a global var with the name of symbol >> >>> >> > and a namespace of the value of the current namespace (*ns*). >> >> >>> >> > In the test below, which succeeds, the var does not appear to end up >> >>> >> > in the "current namespace" per this definition. Am I misinterpreting >> >>> >> > something, or is this a deviation from the spec/reference? >> >> >>> >> > (ns learn.clojure.test.core >> >>> >> > (:use [clojure.test])) >> >>> >> > (deftest namespace2 >> >>> >> > (in-ns 'my.new.namespace) >> >>> >> > ;confirm the current namespace >> >>> >> > (is (= "my.new.namespace" (str *ns*))) >> >>> >> > ;attempt to def a var in the current namespace >> >>> >> > (def anything 10) >> >>> >> > ;the var is not defined in the current namespace >> >>> >> > (is (nil? (ns-resolve *ns* 'anything))) >> >>> >> > ;the var is however definined in the orginal namespace >> >>> >> > (is (not (nil? (ns-resolve (find-ns 'learn.clojure.test.core) >> >>> >> > 'anything)))) >> >>> >> > (is (= 10 learn.clojure.test.core/anything))) >> >> >>> > -- >> >>> > 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 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 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 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
