Rounding errors

[Rune Zedeler]

Perhaps I should wait with this until after 1.4, but, the hell, I'll
mention it now and then we'll go back to it in 1.5 :)

It is a shame that lilypond produces those small rounding errors on the
grobs positions.
- On our ps-printers the system-start-delimiter gets a bit too short. It
is correctly alligned on the top, but the bottom does not quite get down
to last staff line. It looks correct in gv but after printout the error
is there.
- I've earlier mentioned the stems' positions on the beams. You try to
end the stem on the center of the beam which is a problem because
scripts are alligned to the stem and not to the beam. The nice solution
would be to allign grobs to beams instead of stems when beams are
present.
- Further more there is a bug that causes the upwards stem on a knee to
go all the way through the beam instead of only to the middle. The
result is that on a from-down-to-up knee the right, upwards stem is
visible below the beam, like this:


              XXXXXX
              XXXXXX
              XXXXXX
              XXXXXX
              XXXXXX
            XXXXXXXX
          XXXXXXXXXX
        XXXXXXXXXXXX  < stem visible below beam
      XXXXXXXX
    XXXXXXXX
  XXXXXXXX
XXXXXXXX
XXXXXX
XXXXXX
XXXXXX
XXXXXX
XXXXXX

- Finally there is lots of small rounding errors, like beams ending one
pixels before/after the stem, etc... Perhaps stems should be made using
direct postscript...?

And a little suggestion for an improvement when looking into the
stem/beam-code:
How about adding a "white beamspace" option making the center stems only
reach the downmost beam, resulting in

 XXXXXXXXXXXXX
 X           X
 XXXXXXXXXXXXX
 X   X   X   X
 X   X   X   X
XX  XX  XX  XX
XX  XX  XX  XX

instead of

 XXXXXXXXXXXXX
 X   X   X   X
 XXXXXXXXXXXXX
 X   X   X   X
 X   X   X   X
XX  XX  XX  XX
XX  XX  XX  XX


-Rune