Re: Implements the 2nd of 5 legs of getting footnotes up and running. (issue4254055)

[Han-Wen Nienhuys]

Hi Mike,

can fix the below points before continuing with pushing stuff for this
patch series?

On Sun, Mar 6, 2011 at 8:46 PM,  <address@hidden> wrote:
> +// ugh...code dup...hopefully can be consolidated w/ above one day

can you make a priority to do this right now?  If it's not done
directly, 'one day' usually does not happen at all.

> +      int pos = orig->spanned_rank_interval ()[LEFT];
> +      Real spanner_placement = min (1.0,
> +                                    max (robust_scm2double
> (me->get_property ("spanner-placement"), -1.0),
> +                                         -1.0));
> +
> +      spanner_placement = (spanner_placement + 1.0) / 2.0;
> +      int rpos = orig->spanned_rank_interval ()[RIGHT];
> +      pos = (int)((rpos - pos) * spanner_placement + pos + 0.5);
> +
> +      if (pos < me->spanned_rank_interval () [LEFT])
> +        return SCM_EOL;

Can you rethink this? The rank numbers are not good measures of
anything.  In particular, adding a new voice/staff with different
rhythmic texture will cause these numbers to fluctuate, and mess up
your positioning.  I still think this is a overall bad idea, and would
love to see examples where users need placement of footnotes, but
placement so inexact that they do not want to use per-item
positioning.  One possibility is to use column->when() as a measure,
as it is least somewhat invariant, but preferably I'd like to scrap
this (especially given the code dup and overall ugliness) until we
have proof that it's really needed.

-- 
Han-Wen Nienhuys - address@hidden - http://www.xs4all.nl/~hanwen