Yes, in my non-Racket-expert view, the core issue is certainly interactions between toplevel and the module system. It seems unfortunate at best that we have: Welcome to DrRacket, version 5.3 [3m]. Language: racket; memory limit: 512 MB. > (define (range lo hi) (print "hi") (if (> lo hi) null (cons lo (range (+ 1 lo) hi)))) > (range 3 7) "hi"'(3 4 5 6) but: Welcome to DrRacket, version 5.3 [3m]. Language: racket; memory limit: 512 MB. > (define range 0) > (define (range lo hi) (print "hi") (if (> lo hi) null (cons lo (range (+ 1 lo) hi)))) > (range 3 7) "hi""hi""hi""hi""hi""hi"'(3 4 5 6 7) > I'm sure this is an old topic, but would there be fewer gotchas if the REPL were in its own module, importing whatever was defined by the definitions window? (I note that range was not defined in Racket 5.2, but we seem to have the same behavior with map, so this is not a new behavior with respect to the top-level.) --Dan <div class="gmail_quote">On Tue, Oct 23, 2012 at 10:02 AM, Jens Axel Søgaard <<a href="mailto:jensaxel@soegaard.net" target="_blank">jensaxel@soegaard.net</a>> wrote: <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> I forgot to add one rationale for having a context dependent meaning of define-values. With the current setup, you can write (module foo some-lang (define (range ...) ...)) without worrying about whether range is defined in some-lang or not. -- Jens Axel Søgaard 2012/10/23 Jens Axel Søgaard <<a href="mailto:jensaxel@soegaard.net">jensaxel@soegaard.net</a>>: <div class="HOEnZb"><div class="h5">> 2012/10/23 Dan Grossman <<a href="mailto:djg@cs.washington.edu">djg@cs.washington.edu</a>>: >> >> Thanks, David. I would be interested in someone walking through how this >> behavior arises -- as well as the design issue regarding the "hopeless >> top-level": Is this the least-bad option available or an unexpected >> consequence? >> >> --Dan > > The expansion in the repl is: > >> (syntax->datum > (expand > '(define (range lo) > (lambda (hi) (if (> lo hi) null (cons lo ((range (+ 1 lo)) hi))))))) > > '(define-values > (range) > (lambda (lo) > (lambda (hi) > (if (#%app > lo hi) > null > (#%app cons lo (#%app (#%app range (#%app + '1 lo)) hi)))))) > > The tricky thing here is that the meaning of a define-values form is > dependent on context: > > In an internal-definition context (see Internal Definitions), a > define-values form > introduces local bindings. > > At the top level, the top-level binding for each id is created > after evaluating expr, > if it does not exist already, and the top-level mapping of each id > (in the namespace linked with the compiled definition) is set to the binding > at the same time. > > An the documentation on an internal-definition context says: > > A define-values form: The lexical context of all syntax objects > for the body > sequence is immediately enriched with bindings for the define-values form. > > <a href="http://docs.racket-lang.org/reference/define.html?q=define-values&q=repl#(form._((quote._~23~25kernel)._define-values))" target="_blank">http://docs.racket-lang.org/reference/define.html?q=define-values&q=repl#(form._((quote._~23~25kernel)._define-values))</a> > > -- > Jens Axel Søgaard > > > -- > -- > Jens Axel Søgaard -- -- Jens Axel Søgaard ____________________ Racket Users list: <a href="http://lists.racket-lang.org/users" target="_blank">http://lists.racket-lang.org/users</a> </div></div></blockquote></div>