Re: Lyrics break estimation of vertical spacing

[Boris Shingarov]

> Looking at Page_breaking::min_page_count(), I am thinking
 > about the correct generalization of the concept of "rod_height",
 > so that we would still have one "current position", but two
 > "current bottom edges" (so the next current position is calculated
 > by juxtaposing the next system's two top edges against these
 > current bottom edges).
 
Or maybe, we could get by with just adjusting the mutual positions
of the two extents when they are unite()'d...
trying to think of what this simplification might break:
ok, why do we care about the extent, with its [UP] and [DOWN]
parts, instead of just the length? - because some of the variables
in the layout are relative to the middle line, not the edges of the
bounding box.  If we vertically shift the begin- and rest- extents
for the systems against each other, we no longer have a clear
definition of middle line, so we are broken.  Am I missing a way
to solve this, in a way which is easier than a full implementation
of two rods?