Re: beam.cc:pure-rest-collision-callback Place on staff lines; issue 2468 (issue 6035053)

[Keith OHara]

On Tue, 24 Apr 2012 23:29:02 -0700, <address@hidden> wrote:

> http://codereview.appspot.com/6035053/diff/16001/lily/beam.cc
> File lily/beam.cc (right):
>
> http://codereview.appspot.com/6035053/diff/16001/lily/beam.cc#newcode1328
> lily/beam.cc:1328: return scm_from_double (0.0);
> On 2012/04/25 06:21:29, Keith wrote:
> > It would seem we do not want an early return in this case; if the beam
> > direction is not yet know, we would prefer to estimate the rest position
> > to returning the wrong value.
>
> What would the "wrong value" be here?

0.0, indicating rest on centerline, is the wrong value.  If there is another 
voice, the first approximation for the rest is to avoid the other voice.  The 
only case where I see it cause trouble is when dots on a rest try to avoid dots 
in another voice, as if the rest needed to stay at centerline.

> http://codereview.appspot.com/5908043/diff/1/lily/beam.cc#newcode1326
> lily/beam.cc:1326: || beam->get_property_data ("direction") ==
> ly_symbol2scm ("calculation-in-progress"))
> On 2012/04/25 06:21:54, Keith wrote:
> > Mike, any idea why you added this?
>
> If this bit isn't there, a circular dependency could be called up later
> on in the function while looking for the direction.

Thanks.  Now I see.  Later on we get_grob_direction(beam) so we can pick the 
head closest to the beam.
I'll move this test down there to protect the get_grob_direction() next time I 
have Linux up.