Templates inherit thread bindings when created

[Raynes]

I ran into a bug today where templates created with deftemplate seem to behave as if created with bound-fn.

Take this example:

(def ^{:dynamic true} *foo*) 
(defn x [] (prn *foo*)) 
(binding [*foo* "foo"] (x))

This has the effect of printing "foo" because when x is called, the binding sets the value of *foo*. If you replace x with a deftemplate, *foo* is unbound and doesn't get the bindings.

I'm actually pretty shocked that I'm the only one who has noticed this after all this time. This makes Enlive pretty hard to use with Noir and Sandbar because their stateful sessions use bindings to reflect the current session. I'm stuck on moving RefHeap to Enlive because I don't want to delve into the major hax that'd be involved in working around this.

Also, is this project going to be actively maintained in the future? I see there are some pull requests and issues that have been festering for a year or longer. They don't really inspire confidence. :p

[cgrand]

fixed by the above commit.

[solussd]

then, in the future, I had a "wtf?" moment with this. :D Thanks for the fix.

[yuanmai]

clone-for still has the same problem:

(deftemplate ...
  (clone-for [x *foo*]
))

I can access foo outside of clone-for, but inside clone-for, foo is always bound to the initial value.