Re: Irregularity in horizontal spacing

[Urs Liska]

Am 21.11.2015 um 00:01 schrieb Simon Albrecht:
> On 20.11.2015 23:19, Urs Liska wrote:
>> -------- Weitergeleitete Nachricht --------
>> Betreff: Re: Irregularity in horizontal spacing
>> Datum: Fri, 20 Nov 2015 23:04:42 +0100
>> Von: Urs Liska <address@hidden>
>> An: address@hidden
>>
>> Am 20.11.2015 um 17:09 schrieb tisimst:
>>> Are you referring to the final 16ths in each of the beamed groups in
>>> the
>>> center stave?
>>>
>> What I mean is: These final 16ths look pretty good, but they cause the
>> corresponding 8th notes in the top and bottom staff to be spaced pretty
>> irregularly. And in effect inacceptably.
>>
>> As I have to get that score ready ASAP I can't wait for a proper fix (if
>> it should be considered a bug) but need a workaround that works better
>> than having to space the notes in all measures manually.
>
> Well, I agree that this is a grave problem, and all the more grave
> since it’s difficult to work around. 

+1
Not spectacular but a serious issue (although probably rather low in
LilyPond's issue categorization ...)

> I now see Trevor could come up with two ways, but both seem to turn
> off optical spacing altogether.

Yes, both are not fully satisfactory. But for my cause the looks with
uniform-stretch is usually significantly better.

Although in some situations the default spacing is actually better.
I've attached another measure where uniform-stretching (first
attachment) causes the stems in the right hand to look really awkward
while the default (second attachment) is way superior but still leaves
the quavers in an acceptable shape.

I have the impression I can't change this along the way by overriding in
the music itself, so I think I'll settle for the uniform-stretch as a
default and tweak individually if necessary.

> It seems like the spacing engine just does not take the quavers into
> account, which would mean making some sort of a compromise between
> optimal spacing for the semiquavers and the quavers. 

I'm not sure if we can even hope to have a machine do that decision in a
reliable manner. So maybe it would be more likely to be successful if we
could provide a property that for example sets an affinity (or whatever
it should be called) to a certain duration. I.e. sth like "prefer
quavers over semiquavers for nice (optical) spacing.

> This is really deep down in the spacing engine and I’m pretty sure
> that it’s unconfigurable. The question is: whom do we have that might
> be able to provide a fix? :-( Jan? Han-Wen? (they’ve rather completely
> left the project, haven’t they?) Mike? Graham? Or David K., are you
> somewhat acquainted with that area of the source?

Maybe Keith would be willing to give a look? (see
http://lilypondblog.org/2014/02/massive-improvement-in-lilyponds-horizontal-spacing/
and the massive improvement Keith achieved in that area)

> At least I’ve created
> <https://sourceforge.net/p/testlilyissues/issues/4665/> – not a
> Defect, but an Enhancement, since it’s about functionality that we
> just don’t have currently…

Thank you.
I've updated it with some more score attachments.

Best
Urs

>
> Yours, Simon

schwan.png
Description: PNG image

schwan.png
Description: PNG image