Re: anonymous functions in define-grobs.scm

[David Kastrup]

"Keith OHara" <address@hidden> writes:

> Near the top of 'define-grobs.scm' it says
> ;;;; WARNING: don't use anonymous functions for initialization.
> and I am wondering why.
>
> I suspected it was to avoid storing a function body in every instance of the 
> type of grob that might have an anonymous function as one of its default 
> properties.
> But then I see that we use
>  (X-offset .
>   ,(ly:make-simple-closure
>     `(,+
>       ,(ly:make-simple-closure
>        (list ly:break-alignable-interface::self-align-callback))
>       ,(ly:make-simple-closure
>        (list ly:self-alignment-interface::x-aligned-on-self)))))
>
> rather than the simpler
>  (X-offset .
>   ,(lambda (g)
>      (+ (ly:break-alignable-interface::self-align-callback g)
>         (ly:self-alignment-interface::x-aligned-on-self g))))
>
> which follows the rule in the WARNING, but does not seem that it would
> reduce the storage needed in each grob.

That's not a rule for avoiding storage: grobs are not copied "deeply",
and if they were, the anonymity of a function would not make a
difference.

My personal guess is that anonymous functions look like garbage and/or
don't print nicely or at all in EG.

Neither does the definition of X-offset using make-simple-closure
(regarding "nicely"), see
<URL:http://lilypond.org/doc/v2.16/Documentation/internals/accidentalsuggestion>.
I mean, ugh.  Possibly the EG generator complains when not finding a
name for a function?

-- 
David Kastrup