Re: What is the problem with "\relative"? (Was: Do we really offer the future?)

[Kieren MacMillan]

Hi Gilles,

>> deprecate \relative, which I now avoid like the plague.
> Why?

1. It doesn’t play well with reuse: both trivial reuse (i.e., cut-and-paste) 
and more advanced (i.e., referenced in variables) require extra care at the 
very least, and outright extra work (e.g., octave checks, transposition, etc.) 
in most cases. This means that sharing bits of music either within a file/piece 
or between files/pieces (or even between users) requires extra work and is 
error-prone. 


2. It makes what should be simple adjustments unnecessarily complicated, with 
unnecessarily large impacts. Consider, as just one example, my paired functions

split =
#(define-music-function (parser location music1 music2)
   (ly:music? ly:music?)
   #{ <<
         { \voiceOne $music1 }
         \context Voice = "2" { \voiceTwo $music2 }
      >> \oneVoice
   #})

splitLU =
#(define-music-function (parser location music1 music2)
   (ly:music? ly:music?)
   #{ <<
         { \voiceTwo $music1 }
         \context Voice = "2" { \voiceOne $music2 }
      >> \oneVoice
   #})

These (and their 3- and 4-voice counterparts) are workhorses in my code, saving 
me endless amounts of typing constructs like

    << { \voiceOne foo } \new Voice { \voiceTwo bar } >> \oneVoice

every time I simply want a short polyphonic section. I could hardly begin to do 
efficient engraving work without them. Now consider the effect of switching, in 
relative mode, from

   b4 \split { a4 } { c,4 }

to

   b4 \splitLU { c,4 } { a4 }

Because relative mode “leaves from” whatever pitch comes immediately before it, 
the first example would output the b followed by a sixth “chord" immediately 
below it (i.e., with the a on top, sitting a second below the b), whereas the 
second example would output the b followed by a third [!!] “chord" (i.e., with 
the c, on top, sitting a seventh below the b). In absolute mode, I am free to 
choose either function (and each is necessary!) at will, without worrying that 
the choice may mess up the outputted pitches.

This same sort of relative shifting happens when you want to switch the order 
of notes as given in a chord, e.g. <c g’ b> outputs a different chord than <c b 
g’>.


3. Many single edits suddenly require two (or more) edits as a result solely of 
relative mode. For example, let’s say you have

    \relative c’ { c d e a }

and you want to change the e to d. Now you must also add an apostrophe to the 
a, to compensate for the relative octave adjustment:

    \relative c’ { c d d a’ }

wasting effort and brainpower (if you even remember to do it the first time, 
rather than compiling before finding the error).


These are only three of the problems. Worst of all, having it is a false 
economy: it’s not actually intuitive for everyone (though it is for me, and was 
right away), as a quick search of the archives will turn up many newbies 
complaining about “how hard it is to remember when to use , and when to use ‘ 
and when to use nothing”.

Yes, it’s a little more work during initial entry of some music to add the 
correct octavation to the pitches. But the majority of pitches I enter fall 
between c, and b’’, so the difference (if any) is minimal. And of course there 
are many patterns which require *more* octavation typing in relative mode than 
absolute, so in those cases absolute mode saves keystrokes.

Hope this helps explain why I don’t use \relative any more, and tell most 
newbies I know to avoid it.

Cheers,
Kieren.

________________________________

Kieren MacMillan, composer
‣ website: www.kierenmacmillan.info
‣ email: address@hidden