Werner LEMBERG <[email protected]> writes:

>>> * A system with more than a single staff.
>>> * One or more staves should have a key signature and no time signature
>>>   (for example, a piece in C major with a clarinet in B-flat).
>>> * One or more staves should have a time signature but no key signature
>>>   (for example, changing from 4/4 to 12/8, or from 2/4 to 6/8).
>> 
>> I think this measure of Britten's War Requiem qualifies, containing:
>> 
>>  * many staves¹
>>  * clef + time signature (B♭ clarinets),
>>  * no clef + time signature (percussion),
>>  * clef + key signature (chorus).
>
> Thanks!  My request was ill-formulated, sorry.  I should have actually
> asked for examples where time signatures in one staff share the same
> horizontal space with key signatures in another staff.  As it turns
> out, the problem affects only pieces with two staves per system:
>
>   https://gitlab.com/lilypond/lilypond/-/merge_requests/3224

Sigh.  No, it doesn't.  The "problem" (as you call the patch in
operation) can occur with any number of staves.  When a staff has a
given element, it is aligned with the same element in all other staves
having that element.

You can perfectly well have one set of staves that have only a time
signature, and another set of staves that have only a key signature, and
then all key signatures will be aligned and all time signatures will be
aligned but they will occupy the same space.

As soon as there is a single staff with both key signature and time
signature, the space will no longer be shared.

> As soon as there are more staves involved, it's rather clear that time
> signatures must always come after key signatures so that they can be
> properly aligned vertically in the score.

Reminds me of the simple rule to figure out where a mathematical proof
goes wrong.  Just look for the first occurence of a phrase like "it is
rather clear", "obviously" or similar.

As long as there is no staff with both key signature and time signature,
everything can be properly aligned vertically in the score without
requiring key signatures to make space for time signatures or vice
versa.

-- 
David Kastrup


Reply via email to