Re: Make sure slurs actually avoid stafflines. (issue 15400049)

[Janek Warchoł]

Hi Keith,

2013/10/23  <address@hidden>:
>
> On 2013/10/23 04:22:22, Keith wrote:
>> The commit message makes it sound like there was a blunder in the
>> direction that
>>
>> the old code moved the slur.  I don't see any such problem.
>>
>> It looks like the old code intentionally moved the middle of the slur
>> always in
>> the direction of the slur, so the curvature would increase.  It simply
>> didn't
>> move far enough, because the curve does not move as far as the control
>> points.
>
> Uh, how had you came to this conclusion?  It's absurd.  Take this
> example:
>
> \relative c'' {
>   c8( b a2) c8( b)
>   \slurUp
>   d,2( f)
>   a2 c
> }
>
> Here's how these slurs look with the correcting function turned off
> altogether - none is crossing the staffline:
> http://lilypond.googlecode.com/issues/attachment?aid=36280006000&name=no-correction.png&token=I4nuTyt1y5gR8lDgjQPipN-8Hm8%3A1382562402092&inline=1
>
> They are just a bit too close to the staffline, that's all.  They should
> be flattened, but the previous version of the algorithm decided that
> such slurs should have their curvature increased, which resulted in
> this:
> http://lilypond.googlecode.com/issues/attachment?aid=36280006001&name=previous-correction.png&token=PeJAER3pHxyZQAqN_3vKl4F2fUY%3A1382562402092&inline=1
>
> Can you tell me in what way this made any sense?  Making the slurs even
> more curved would make them even more ugly.
>
> Also, please look at the example in our essay:
> http://lilypond.org/doc/v2.17/Documentation/essay/automated-engraving#improvement-by-benchmarking
> Note that Baerenreiter puts the slur in the 2nd measure under the
> staffline, just as i want it to be.

I hope that you won't interpret the harsh tone of my email as a personal attack.
I spent *8 hours* (measured with a watch) reading slur code and
working on this patch.  I got irritated when i saw your email, because
it seemed to me that when i said that something was broken you didn't
trust my words enough to double-check your doubts.

I realize that you probably didn't mean it.  Unfortunately this
resonated with some of my other experiences from LilyPond development,
when
a) my questions quite often didn't get answered desipte being short
and well-formulated,
b) my plain, well-commented and carefully git-crafted patches
sometimes don't get reviewed for months,
c) my diagnoses and observations are quite often not trusted without a
lot of disussion and necessity to provide lots of scanned examples,
d) it seems that people generally don't care or notice things that i
care about and notice (for example when i say "many slurs, beams and
ties in LilyPond are really ugly" many people don't believe me - as if
they trusted LilyPond's decisions more than my judgement - or at least
so it seems to me).

This is all really unfortunate; of course you're not responsible for
this situation.  But it made me frustrated nevertheless, so i decided
to explain myself.

Anyway, be sure that i *do appreciate* getting reviews, and i'm
honored to see Han-Wen reviewing my patch!

good night
Janek