bug#10246: R6RS library form broken?

bug#10246: R6RS library form broken?

[Tobias Brandt]

Hi,

I'm running the latest git version (2.0.3.244-04ec29).
Declaring a R6RS style library doesn't work:

(library (foo))

=>
ice-9/psyntax.scm:1422:30: In procedure expand-macro:
ice-9/psyntax.scm:1422:30: Syntax error:
source expression failed to match any pattern in form (library (foo))


while this does:

(define-module (foo))

bug#10246: R6RS library form broken?

[Ian Price]

Tobias Brandt <tob.brandt@googlemail.com> writes:

> Declaring a R6RS style library doesn't work:
>
> (library (foo))
>
> =>
> ice-9/psyntax.scm:1422:30: In procedure expand-macro:
> ice-9/psyntax.scm:1422:30: Syntax error:
> source expression failed to match any pattern in form (library (foo))
>
>
> while this does:
>
> (define-module (foo))

This is a misunderstanding on your part, I'm afraid. R6RS library forms
are intended to wrap the code[0]; they are not declarations to be placed at
the top. The manual does mention this[1], by noting the syntax for library
is

(library name (export export-spec ...) (import import-spec ...) body ...)

Without an example, I can see how the equivalence mentioned at the top
may be misleading. Perhaps the manual should add an example? Even
something as simple as, say

(library (counters)
(export make-counter)
(import (rnrs base)

(define (make-counter start)
  (lambda ()
    (let ((val start))
      (set! start (+ start 1))
      val)))

)

would probably be helpful.

0. http://www.r6rs.org/final/html/r6rs/r6rs-Z-H-10.html
1. http://www.gnu.org/software/guile/manual/html_node/R6RS-Libraries.html

-- 
Ian Price

"Programming is like pinball. The reward for doing it well is
the opportunity to do it again" - from "The Wizardy Compiled"

bug#10246: R6RS library form broken?

[Andy Wingo]

On Thu 08 Dec 2011 00:00, Ian Price <ianprice90@googlemail.com> writes:

> Tobias Brandt <tob.brandt@googlemail.com> writes:
>
>> (library (foo))
>
> This is a misunderstanding on your part, I'm afraid. R6RS library forms
> are intended to wrap the code[0]; they are not declarations to be placed at
> the top. The manual does mention this[1], by noting the syntax for library
> is
>
> (library name (export export-spec ...) (import import-spec ...) body ...)

Indeed.  Thanks for the answer, Ian.

In the new email-happy bug system, to close a bug you edit the Cc: list
to add -done after the bug number, as I have done here.  Feel free to do
this for any bug that you feel is done; if we make a mistake, we can
simply re-open it.

I'm sure we could do better accounting as to the kind of resolution and
such, but I'm most concern with simply marking bugs as being resolved.

Regards, and happy hacking to you both :-)

Andy
-- 
http://wingolog.org/