Hesitant Steps Forward

So, I just dived in and started trying to implement duration as a role.

role ABC::Duration {
    has $.ticks;

    our multi sub duration-from-parse($top) is export {
        ABC::Duration.new(:ticks($top.Int || 1));
    }

    our multi sub duration-from-parse($top, $bottom) is export {
        if +($top // 0) == 0 && +($bottom // 0) == 0 {
            ABC::Duration.new(:ticks(1/2));
        } else {
            ABC::Duration.new(:ticks(($top.Int || 1) / ($bottom.Int || 1)));
        }
    }

    our method Str() {
        given $.ticks {
            when 1 { "---"; } # for debugging, should be ""
            when 1/2 { "/"; }
            when Int { .Str; }
            when Rat { .perl; }
            die "Duration must be Int or Rat, but it's { .WHAT }";
        }
    }
}

The corresponding action for an explicit note length is

    method note_length($/) {
        if $<note_length_denominator> {
            make duration-from-parse($<top> ?? $<top>[0] !! "", $<note_length_denominator>[0]<bottom>[0]);
        } else {
            make duration-from-parse($<top> ?? $<top>[0] !! "");
        }
    }

I messed around a bunch trying to make "e" does Duration work as a type, but eventually gave up and just coded an ABC::Note type:

class ABC::Note does ABC::Duration {
    has $.pitch;
    has $.is-tie;

    method new($pitch, ABC::Duration $duration, $is-tie) {
        say :$duration.perl;
        self.bless(*, :$pitch, :ticks($duration.ticks), :$is-tie);
    }
}

with corresponding action

    method mnote($/) {
        make ABC::Note.new(~$<pitch>,
                           $<note_length> ?? $<note_length>[0].ast !! ABC::Duration.new(1),
                           $<tie> eq '-');
    }

So… that seems to work okay so far. But it does raise some issues for me.

1. I’m finding this $<top> ?? $<top>[0] !! "" pattern to be very repetitive. Surely there must be a better way to do it? (errr… wait a minute, could that be just $<top>[0] // ""?)

2. I don’t mind the proliferation of small classes in my code, that corresponds nicely to what we are modeling. But I am starting to mind the corresponding proliferation of small source files. Is there a better way to organize things?

… and that’s all I can remember at the moment. I think I’ll wander off and have breakfast.

About these ads