Re: Fixes issue 39 by raising stems (issue3934041)

[Mike Solomon]

Thanks Keith,

I think that the case you're talking about (<< e'32. \\ e'2>>) is a problem 
with my code, which triggers stem raising for flags that fall on the right even 
though these flags do note cause intersection problems.  Is there a good way to 
weed this out?

On Jan 13, 2011, at 1:03 AM, address@hidden wrote:

> On 2011/01/12 15:34:22, MikeSol wrote:
>> Fixed (I think).
> 
> I like how this patch behaves!
> No problems at all this time in the music I threw at it  (but I haven't
> yet seen the original issue 39 in music either).
> 
> Playing around with the various situations, only 2 problems.
> 
> 1) the flag in << e'32. \\ e'2>> raises enough to let the dot try to
> slip in, and the dot merges with the flag.  Adding 0.2 to raise_tip
> gives enough clearance.  This case isn't really within issue39 because
> the flagged note goes on the right, but it falls within the test I
> suggested:
>> > I think the ideal target set is (touch && !merge_possible) but I
> will be slow
> 
> 2) the flag raises for no reason on a few chords where stem-shorten
> affects the flag; compare
> << <a' c''>16 \\ a'2 >>  << <c'' e''>16 \\ c''2 >>
> 
See below.
> 
> http://codereview.appspot.com/3934041/diff/29001/lily/stem.cc
> File lily/stem.cc (right):
> 
> http://codereview.appspot.com/3934041/diff/29001/lily/stem.cc#newcode398
> lily/stem.cc:398: if (initial_length + extra_stem_length + hp[DOWN] >
> stem_end)
> stem-shorten fooled this filter by reducing your pre-computation of
> stem_end, but not reducing initial_length.  I see the contortions you
> went through to apply your correction within stem_length().  If you
> think it fits more naturally in calc_stem_end_position, that is probably
> wiser.  I think I see a good fit but this time I'll test before I
> suggest.
> 
I am working on an article for the next two days, so if you have time to cook 
up a patch that better deals with stem-shorten, I'd be much obliged!

Cheers,
MS

> http://codereview.appspot.com/3934041/