Re: Issue 1471 in lilypond: accidentals after a clef change are not printed

[lilypond]

Comment #4 on issue 1471 by address@hidden: accidentals after a clef change  
are not printed
http://code.google.com/p/lilypond/issues/detail?id=1471

Ok, here is the full job which passes the regtest without difference in  
output, but produces the following image for the test file of this bug  
report.

There are several downsides to this patch: the most important one likely  
being that it uses +nan.0 for signifying an unspecified alteration.   
Replacing this by 1000 had the unexpected result of causing extra naturals  
upon a clef change.  +nan.0, apart from being more appropriate, has no  
discernible sign, so it does not suffer from that defect.

Using #f lead to errors because alterations are compared using `='.  The  
code became more complex than expected because localKeySignature is getting  
_stuffed_ with non-altered pitches as well as altered ones.  If one does  
not leave them untouched, one gets nonsensical cautionary accidentals after  
clef changes.  Weeding them out altogether will change the accidental  
behavior, likely to the behavior that this strange measure is trying to fix  
in the first place.

Attachments:
        0001-Issue-1471-Invalidate-alterations-upon-key-change-ra.patch  3.2 KB
        junk.preview.png  2.9 KB