Re: clef change confuses manual key signature

[David Kastrup]

james <address@hidden> writes:

> On Aug 15, 2012, at 9:54 AM, David Kastrup wrote:
>
>> Keith OHara <address@hidden> writes:
>> 
>>> David Kastrup <dak <at> gnu.org> writes:
>>> 
>>>> That image does not make sense to me at all.  Notes appear in key 
>>>> signature (though in a different octave) and still carry an accidental. 
>>>> How do you distinguish a normal key signature (valid across all octaves) 
>>>> from a restricted-octave one (valid only in one octave)?  They look the 
>>>> same.
>>> 
>>> Lilypond docs do not seem to explain any way to print the key signature
>>> accidentals on different lines than standard, except for this crazy method
>>> where the alterations count for just one octave.
>>> 
>>> <speculation>
>>> There was no way to alter the printing of the key signature, 
>>> someone needed to do so, found the data structure for the local key 
>>> signature that tracks transient accidentals, including octave, used
>>> that as a way to serve his need, and posted to the snippets list.
>>> <end speculation>
>> 
>> And composers all over the land adopted "this notation".  Sounds like a
>> Microsoft success story.
>> 
>>> It would be better to use standard key signatures with custom scales
>>>  wholetone = #`((0 . ,NATURAL) (1 . ,NATURAL) (2 . ,NATURAL)
>>>      (3 . ,SHARP) (-3 . ,NATURAL) (-2 . ,FLAT) (-1 . ,FLAT) )
>>>  { \key d\wholetone  bes1 }
>>> and adapt the print routine
>>>  key-signature-interface::alteration-position
>>> to allow for more flexible printing.
>> 
>> No idea.  At any rate, I am going for the "valid in all octaves even if
>> octave is given" angle.  Of course that is incompatible with current
>> behavior, but current behavior is incompatible with common sense or
>> logic.  It is not even possible to guess the pitches one is supposed to
>> play.
>
> Honestly, I don't know what the original intent of lilypond's behavior
> was supposed to be.

I consider Keith's theory that somebody poked LilyPond internals with a
stick and made a snippet from the resulting thrashing quite plausible.

LilyPond's code clearly has not been written with the intent of
supporting octave-fixed signatures.

-- 
David Kastrup