Re: [Lilypond-auto] Issue 2990 in lilypond: \RemoveEmptyStaves in StaffGroup context crashes

[lilypond]

Comment #16 on issue 2990 by address@hidden: \RemoveEmptyStaves in StaffGroup  
context crashes
http://code.google.com/p/lilypond/issues/detail?id=2990

Correct me if I am wrong, but it would appear that the code crashing on  
cyclical element lists is the skyline code.  It is a simple matter of  
storing 'cycle in the elements field you are recursing through temporarily  
to avoid infinite recursion.  We do this with grob dependencies elsewhere.   
Now the "elements" situation usually also corresponds to an axis parent  
group situation, and we don't do this sort of cycle-breaking when  
determining parents.  Why don't we get parenting cycles but elements  
cycles?  It would appear that we are doing something wrong here.

You are very eager to paste over the actual problem.  Why?  Why not figure  
out what is wrong here and do a proper fix making use of those parts of the  
data and relations we can (and do elsewhere) work with efficiently?

I really resent planless scattering of expensive bandaids everywhere.  That  
results in an incoherent mess of patched-together half-working code nobody  
will be able to maintain since it is totally unclear what code is actually  
effective and what code has its effects overriden elsewhere.