Boris wrote: > Right now, we start from the list of markups, and for-each markup, > apply the markup function and destructively append! the resulting > stencil to the result list under construction. > > The procedural nature of this design, is somewhat limiting in terms of > expressive power. For what I am trying to do -- building a more > flexible/general markup formatting facility -- the formatting of a > markup may sometimes be context-controlled, i.e. formatting of the next > piece of markup may depend on how the previous markups were formatted. For > this, the design where the markup's stencil is completely defined > by the single markup element in the list fed into > interpret-markup-list, is not sufficient. One way to address it, is > simply to pass another argument to the markup > procedure, containing the context where we are along the way. A more > elegant way is to do it in continuation passing style: the markup > function, which represents one step in building of the stencils list, > takes the tail of the markup list (the head is the current markup > function itself) and the partially-built list of stencils, formats the > current piece pf markup, tucks it on to to the stencils-list being > built, and tail-recursively executes the rest of the markup with list > of stencils grown by the just-now-formatted stencil. This way, each > step can reason about what has been already built, looking backward > (although only in terms of stencils), and what is yet to be built, > looking forward (although only in terms of markups).
I'm not sure what you have in mind here, but I would agree with this being a more elegant way of doing it. I always look at ! functions in Scheme as something that would be best avoided if possible, because there's generally a more expressive way to do things that avoids the ! function. Thanks, Carl _______________________________________________ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
