unofficial mirror of bug-guile@gnu.org 
 help / color / mirror / Atom feed
* bug#16726: Possible bug?
@ 2014-02-12  0:51 Nigel Warner
  2014-02-12  3:40 ` Mark H Weaver
  0 siblings, 1 reply; 2+ messages in thread
From: Nigel Warner @ 2014-02-12  0:51 UTC (permalink / raw)
  To: 16726

#|
   This file is an extract of a documentation and code assembly process
   which works from a list of elements and attributes. Perhaps I'm
   being particularly stupid but I can't see why the function
   emit-groff-bug? does not work as expected.

   Version is 2.09 GCC is 4.7.3 on 64 bit Ubuntu self compiled.

   BTW, thanks to all the developers for a wonderful program. I am
   greatly enjoying programming in Guile.

|#
(define-syntax say
   (syntax-rules ()
    ((say format-string ...)
     (format #t format-string ...))
    ((say format-string)
     (format #t format-string))))

(define-syntax emit
   (syntax-rules ()
     ((emit a-string)
      (say (string-append a-string "\n")))
     ((emit a-string ...)
      (say (string-append (format #f a-string ...) "\n")))))


(define canvas-description
"A canvas object specifies the two dimensional area on which graphics
objects are painted.

The function `canvas' requires at least one parameter, the name by
which the canvas will be known. If there is already a canvas by that
name in the canvas-table then it will be used as the current
canvas. If not then a new canvas will be created using the default
values unless overridden by keyword value pairs passed to the
function.")

(define thoth-canvas
   `((#:object-name . canvas)
     (#:description . ,canvas-description)
     (#:attributes
       (viewbox none "Creates user coordinate space")
       (background none "Creates a background rectangle")
       (transform none "Transformation object"))))

(define (process-attributes-for-groff lst)
   (for-each
     (lambda (element)
      (emit ".attribute ~a ~a" (list-ref element 0)
                               (list-ref element 1))
      (emit ".start-hint")
      (emit "~a" (list-ref element 2))
      (emit ".end-hint"))
    lst))

(define (emit-groff lst)
   (for-each
    (lambda (element)
      (cond ((and (pair? element)
                  (eq? #:object-name (car element)))
             (emit ".object-name ~a" (cdr element)))
            ((and (pair? element)
                  (eq? #:description (car element)))
             (emit ".start-description")
             (emit "~a" (cdr element))
             (emit ".end-description"))
            ((and (list? element)
                  (eq? #:attributes (car element)))
             (process-attributes-for-groff (cdr element)))
            (else
             (emit "emit-groff | unknown list type")
             (throw 'snafu))))
    lst))

(define (emit-groff-bug? lst)
   (for-each
    (lambda (element)
      (cond ((pair? element)
             (cond ((eq? #:object-name (car element))
                    (emit ".object-name ~a" (cdr element)))
                   ((eq? #:description (car element))
                    (emit ".start-description")
                    (emit "~a" (cdr element))
                    (emit ".end-description"))))
            ((list? element)
             (cond ((eq? #:attributes (car element)))
                   (process-attributes-for-groff (cdr element)))
             (else
              (emit "emit-groff | unknown list type")
              (throw 'snafu)))))
    lst))

  (emit "------------- working")
  (emit-groff thoth-canvas)
  (emit "------------- not working")
  (emit-groff-bug? thoth-canvas)






^ permalink raw reply	[flat|nested] 2+ messages in thread

* bug#16726: Possible bug?
  2014-02-12  0:51 bug#16726: Possible bug? Nigel Warner
@ 2014-02-12  3:40 ` Mark H Weaver
  0 siblings, 0 replies; 2+ messages in thread
From: Mark H Weaver @ 2014-02-12  3:40 UTC (permalink / raw)
  To: Nigel Warner; +Cc: 16726-close, request

tags 16726 notabug
thanks

Hi,

Nigel Warner <nwarner84@gmail.com> writes:
>   This file is an extract of a documentation and code assembly process
>   which works from a list of elements and attributes. Perhaps I'm
>   being particularly stupid but I can't see why the function
>   emit-groff-bug? does not work as expected.

In the future, please show us what output the program produced,
and what output you expected.

There are several mistakes in your program.

> (define-syntax say
>   (syntax-rules ()
>    ((say format-string ...)
>     (format #t format-string ...))
>    ((say format-string)
>     (format #t format-string))))

The second rule will never be used, because the first rule covers that
case.  In macros, "x ..." matches zero or more items.

> (define-syntax emit
>   (syntax-rules ()
>     ((emit a-string)
>      (say (string-append a-string "\n")))
>     ((emit a-string ...)
>      (say (string-append (format #f a-string ...) "\n")))))

This is a bad idea, because you are conflating the format string with
the strings you want to output.  If the string you want to output
includes "~", then it will misbehave.  Did you notice the compiler
warnings about non-literal format strings?

> (define (process-attributes-for-groff lst)
>   (for-each
>     (lambda (element)
>      (emit ".attribute ~a ~a" (list-ref element 0)
>                               (list-ref element 1))
>      (emit ".start-hint")
>      (emit "~a" (list-ref element 2))
>      (emit ".end-hint"))
>    lst))

I suggest that you use the (ice-9 match) module to destructure 'element'
into its components.  The code will look a lot nicer, and it will also
be more efficient.

> (define (emit-groff lst)
>   (for-each
>    (lambda (element)
>      (cond ((and (pair? element)
>                  (eq? #:object-name (car element)))
>             (emit ".object-name ~a" (cdr element)))
>            ((and (pair? element)
>                  (eq? #:description (car element)))
>             (emit ".start-description")
>             (emit "~a" (cdr element))
>             (emit ".end-description"))
>            ((and (list? element)
>                  (eq? #:attributes (car element)))
>             (process-attributes-for-groff (cdr element)))
>            (else
>             (emit "emit-groff | unknown list type")
>             (throw 'snafu))))
>    lst))
>
> (define (emit-groff-bug? lst)
>   (for-each
>    (lambda (element)
>      (cond ((pair? element)
>             (cond ((eq? #:object-name (car element))
>                    (emit ".object-name ~a" (cdr element)))
>                   ((eq? #:description (car element))
>                    (emit ".start-description")
>                    (emit "~a" (cdr element))
>                    (emit ".end-description"))))
>            ((list? element)
>             (cond ((eq? #:attributes (car element)))
>                   (process-attributes-for-groff (cdr element)))
>             (else
>              (emit "emit-groff | unknown list type")
>              (throw 'snafu)))))
>    lst))

There are several mistakes in this procedure:

* Every list is also a pair, so the (list? element) case is never taken.
  The (pair? element) case is taken for the #:attributes, and then no
  matching case is found in the inner 'cond' there.

There are two parenthesis errors:

* You have the (process-attributes-for-groff (cdr element)) in a
  position where its treated as a clause of the inner 'cond', instead of
  as an expression within the (eq? #:attributes (car element)) case.

* You've put the 'else' within the "(list? element)" case, instead of as
  its own clause.  Did you see the compiler warning about "possibly
  unbound variable `else'"?

     Mark





^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2014-02-12  3:40 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2014-02-12  0:51 bug#16726: Possible bug? Nigel Warner
2014-02-12  3:40 ` Mark H Weaver

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).