takes 100ms.

> While it would of course be ideal for our compiler to efficiently
> handle expressions 10000 levels deep (if it can be done without
> sacrificing maintainability), dealing with such pathological cases
> should not be a high priority, IMO.  There are many more important
> things to work on.
>
> Is this just an academic exercise, or does Lilypond generate massively
> huge expressions like this?

LilyPond evaluates a lot of one-shot expressions in the course of its
operation, including complex ones.  But Guilev2 offers enough other
stumbling blocks.  We have not yet arrived at a state where it would
even be possible to evaluate performance.

I tripped over this problem here merely while trying to find a
persuasive example for benefits of improving scm_reverse_x performance,
as scm_reverse_x is used quite a bit in libguile/expand.c.

Since the performance impact of reverse! in this example is
indistinguishable from thermal noise, its use seems to be outside of the
quadratically behaving code parts.

-- 
David Kastrup