Re: wrong beam positions in LilyPond

[Janek Warchoł]

On Fri, Aug 31, 2012 at 5:58 PM, Han-Wen Nienhuys <address@hidden> wrote:
> On Fri, Aug 31, 2012 at 8:24 AM, Janek Warchoł <address@hidden> wrote:
>> i can confirm that the beaming remained virtually
>> unchanged since LilyPond 2.6 (see attached "old vesions").
>> I've also checked LilyPond 2.16, and in some cases the beaming is
>> better, but many problems remain.
>
> That is a testament to the regtest's usefulness, and it was certainly
> intended to be like that.

:)

>> Would you be interested in fixing beaming code together?
>
> Sure,
>
> fixing things is always good. The beaming code is pretty generic, so I
> expect that this will mostly be about tweaking scoring functions, and
> making sure existing situations stay the same.

On Fri, Aug 31, 2012 at 8:47 PM, Jan Nieuwenhuizen <address@hidden> wrote:
> I guess it's only obvious once you see it, and Janek prepared
> very nice and intstructive examples, for easy comparison with Ross.

thanks :)  I'm currently in the process of revising them.
How shall we organise our discussion?  I guess that dumping all ~500
beaming examples somewhere and having a spontaneous discussion on how
they should look like wouldn't make much sense.
During my analysis i've identified several issues relevant to many bad
beams, together with possible solutions.  What about me introducing
these ideas one by one (Graham-GOP-style)?  After we decide on them,
we'll take a look at all my examples together and see whether the
results are satisfactory.


On Fri, Aug 31, 2012 at 8:53 PM, address@hidden
<address@hidden> wrote:
> You can set debug-beam-scoring = ##t in the paper bloc.

thanks!  Is there any explanation to these numbers other than digging
it myself from beam-quanting.cc?  I've skimmed over it, but didn't
find it immediately helpful.

cheers,
Janek