----------
Pitfall #1
----------

Consider this code in a render method:

(iter (for newval in '("a" "b" "c"))
      (let ((newval-1 newval))
        (<ucw:a :action (setf (value test) newval-1)
          (<:as-html newval))))

Seems like it should work fine, the variable is even rebound, but in fact it
won't work due to something that can be considered a bug, but a non-trivial to
fix one.

segv            it's iterates 'fault'. iterate walks and macroexpands its body but
                doesn't modify the lexical environment while it does that, so it causes with-call/cc
                to get expanded with the iter's env, and not the iter's env + the let binding.
attila_lendvai  ehh
segv            (fwiw arnesi's own code walker doesn't do that either)
segv            this explains why it works fine with loop and dolist
segv            and dolist*
attila_lendvai  is it easy to fix? i can push to iterate, but have no experience with envs
segv            attila_lendvai: i requires looking through the various lisps to figure out how to
                do that, and fixing iterate's code-walker to do the proper environment augmentation
                for let, let*, flet, labels, macrolet and symbol-macrolet forms.
segv            doable, but not trivial.
segv            at that point it'd probbaly be easier to rework iterate to use arnesi's code walker
                (which would be easier to fix since some of the non portable environment stuff is already there)
attila_lendvai  hm, doesn't sound like something i would chop my axe in... :)
attila_lendvai  iterate is non-trivial in itself
segv            yup.

Workarounds: 
  - use loop or dolist until iterate is fixed.
  - in the ajax branch you can disable with-call/cc wrapping (which is not needed here):
    (iter (for newval in '("a" "b" "c"))
          (let ((newval newval))
            (<ucw:a :action (register-action () ;; :with-call/cc is #f by default
                              (setf (value self) newval))
                    (<:ah newval))))