Re: book-predicate

[David Kastrup]

Jan-Peter Voigt <address@hidden> writes:

> sorry for this kind-a-basic-scheme-question(s) ... ;-)
> How do I inject variables into an empty module?
>
> I assume, the string-eval approach loads up the whole
> guile-tool-chain, so that define is defined.
> The function I am thinking of shall work like this:
>
> <pseudo-snip>
> #(define-public (set-book-headers! book header) ; book is a book,
> header is an a-list, containig all vars to inject into header
>   (let ((bookhead (ly:book-header book)))
>        ; if book has no header, create one
>        (if (not bookhead) (let ((bh (/make-module/)))
>                 (set! bookhead bh)
>                 (ly:book-set-header! bookhead)
>        ))
>        (for-each (lambda (p)
>                          (if (pair? p)
>                              (let ((key (car p))
>                                    (val (cdr p)))
>                                   (/inject key=val/ bookhead))))
>                  header)
> ))
> </pseudo-snip>
>
> So AFAICS there are at least 3 different ways to create a module to
> use in a header:
> 1: (eval-string "(define-module (a b))")

What's with the eval-string nonsense?  Why don't you just use
(define-module (a b)) instead?

> 2: (ly:book-header #{ \book { \header { } } #})
> 3: (make-module)
>
> The first two are some kind of cheating, but create modules, where I
> can inject my vars using
> (eval `(define ,key (quote ,val)) bookhead)
>
> I would prefer the 3rd one, if I knew, how to inject the vars inside
> the empty module.

(module-define! module key val)

Not all that hard.

> In my first post, regarding this stuff, I used eval-string, because I
> assume it produces less overhead than creating a book in inline
> lily-code.

Sure, but why eval-string?  Why not write what you want instead?

-- 
David Kastrup