Re: [dssslist] in reply to: "define inside let body"

[Joaquín Zuazo <jz@xxxxxxxxx>]

Thanks to all.
In my opinion, according to the standard, this simple an unuseful
example should work:

(define test-variable
	(let (( void-variable 0))
	  (define a 2)
	  (define b 2)
	  (define c 2)
	  ;; all defines you want to add here are local variables
	  ;; inside let

	  ( + a b c)))

Simply, Openjade does not at the moment work with this construction,
like tmcd has pointed, and with his help I have seen also in the
documentation " Actual limitations of Openjade".
In any case, learning form the available documentation is hard an
perhaps, for me, harder. I have worked with common lisp and emacs lisp
and with dsssl things are different from what I am used to do.

I'll try to get the Farrere's book. I appreciate your opinions about it.
I have found it in some stores in the web and I have read the table of
contents. There is said the book contains about 340 pages and I have the
doubt if it is enough to explain with clarity and with examples all the
difficulties of the specification. The standard document has 293 and is
quite full of black letters. You can write to my address if you think
some rule is broken. In any case, congratulations to Javier for having
this book written, after looking at the examples of this mail list, and
the docbook's styles sheets code any grain of sugar for this cake sounds
nice.

El miC), 09 de 02 de 2005 a las 20:00, tmcd@xxxxxxxxx escribiC3:
> > >Hi, I am trying to understand the dsssl standard document
> > >(ISO/IEC 10179:1996(E). Trying to put this example to work with
> > >openjade, part 8.4 "Definitions" of the document, at top level:
> > >
> > >(define p
> > >   (let ((x 5))
> > >     (define foo (lambda (y) (bar x y)))
> > >     (define bar (lambda (a b) (+ (* a b) a)))
> > >     (foo (+ x 3)))
> > >)
> >
> > Hi,
> > I'am rather inexperienced but I'll try to answer.
> > The former should look like:
>
> The example above is right out of the DSSSL spec, page 43.  I don't
> have a net connection right now, btu I suspect that Googling for
> "dsssl96b.pdf" would find it.  Look for "dsssl96f.pdf" too: it's the
> table of contents.
>
> > (define p
> >    (let ((x 5)
> >      (foo (lambda (y) (bar x y)))
> >      (bar (lambda (a b) (+ (* a b) a))))
> >      (foo (+ x 3)))
> > )
> >
> > It won't work anyway because of recursive definitions, but now at
> > least it should be accepted by jade.
>
> See "NOTE 9" quoted below.  I'm not sure that it won't work, but I
> haven't tried it.  In any event, "letrec", which was mentioned as the
> equivalent expansion, makes it work.
>
> > >According the standard. body is described (in the same page) as
> > >
> > > [69] body=definition* expression
> >
> > In your code you repeat expression part 3 times, so it is not very
> > correct.
>
> The BNF listed on pages 42 and 43 are
>
>     [66] definition = variable-definition | procedure-definition
>
>     Definitions may take two possible forms.
>
>     [67] variable-definition = (define variable expression)
>
>     This syntax is primitive.
>
>     [68] procedure-definition = (define (variable formal-argument-list)
body)
>
>     This form is equivalent to
>
>     (define variable
>         (lambda (variable formal-argument-list) body)).
>
>     A definition that does not occur within an expression is known as
>     a top-level definition.
>
>     A top-level definition
>
>     (define variable expression)
>
>     evaluates expression in the top-level environment and binds
>     variable to the result in the top-level environment.
>
>     The expression in a top-level definition shall not be evaluated
>     until all top-level variables that would be referenced by
>     evaluating the expression have been defined.
>
>     NOTE 9 This constraint does not prevent the definition of mutually
>     recursive procedures, because evaluating a lambda expression does
>     not reference variables that occur free within it.
>
>     It shall be an error if it is impossible to evaluate all the
>     expressions occurring in top-level definitions in such a way that
>     this constraint is not violated. ...
>
>     [69] body = definition* expression
>
>     Definitions may also occur at the beginning of a body. These are
>     known as internal definitions. The variable defined by an internal
>     definition is local to the body. The region of the binding is the
>     entire body. For example,
>
>         (let ((x 5))
>             (define foo (lambda (y) (bar x y)))
>             (define bar (lambda (a b) (+ (* a b) a)))
>             (foo (+ x 3)))          =>   45
>
> So in that example, "definition" is repeated 2 times:
>
>     1: (define foo (lambda (y) (bar x y)))
>     2: (define bar (lambda (a b) (+ (* a b) a)))
>
> and expression appears once:
>
>      (foo (+ x 3)))
>
> It's a bit of a pity that OpenJade doesn't allow it: I could have
> gotten rid of a few "let" blocks that way.  If it allowed definitions
> before any _expression_ (extending the standard), a LOT of "let"
> blocks would go away.  Still, it's easy enough to work around.