Re: Issue 1110 in lilypond: Wrong octave of repetition chord with \relative and #{ #} syntax

[lilypond]

Comment #32 on issue 1110 by address@hidden: Wrong octave of repetition chord  
with \relative and #{ #} syntax
http://code.google.com/p/lilypond/issues/detail?id=1110

The dilemma basically is that "last chord" is not well-defined.  \relative  
has an order for "last pitch" that it follows.  The parser has an order  
for "last default duration" that it follows.  Both are different but  
thankfully orthogonal.  "last chord" is not.  It gets garbled when  
\relative does not know which chords belong together, so it makes sense to  
define "last chord" like \relative does.  Unfortunately we don't know when  
and whether at all \relative will run on the material, so a \relative-only  
answer to the order is not going to fly.

However, time order (like shown with parallel music which is _not_ two  
voices) is too different from what LilyPond does elsewhere to make much  
sense.  It would be very easy to implement.

Is
input/regression/repeat-tremolo-chord-rep.ly
a corner case?  We have here
\relative c' {
  <c e g>1
  \repeat tremolo 4 q16
  \repeat tremolo 4 { q16 }
  \repeat tremolo 4 { c16 q16 }
}
Before 2240,we _had_to involve the parser in the "last chord" concept  
because only the parser had the information which EventChord would even  
count as a chord.  Personally, I think a single run through the whole score  
when iteration _starts_ is probably the most useful place.  Forcing the  
user to be explicit by doing last chord substitution when called with a  
command is even more predictable, but inconvenient.