Re: \relative is not the best way of entering complicated music

[David Kastrup]

Kieren MacMillan <address@hidden> writes:

> Hello all,
>
> Just thought I'd put my 2¢ in here…
>
> In the last year alone, I have Lilypounded two stage musicals, a
> chamber opera, and about a dozen smaller pieces, with a range of
> styles from Alan Menken to Arnold Schoenberg. So I believe I can
> qualify as someone who has entered "complicated music".
>
> For me, \relative is unquestionably the best way of entering
> music. More specifically, \relative c'' { c ... } has been the best
> way of entering music for me. I find it totally intuitive, both in
> terms of how to enter it, and what I should expect Lilypond to
> output. When combining variables — consecutively or concurrently — it
> does exactly what I expect and want.
>
> As far as I can see, the new syntax proposal (i.e., \relative { c'' …
> } where c'' is an absolute reference) provides me with the same level
> of intuitive simplicity, so I am fully in support of the proposal.

Well, three points:
a) you can stay with your current usage easily enough
b) letting convert-ly change all of your existing files without explicit
   instruction would seem inappropriate
c) in \relative { c'' ..., the pitch might not actually be c''.  If you
   are wired to the reference pitch c'', having to think about whether
   you need to write \relative { a' } or \relative { a'' } instead of
   \relative c'' { a } might be unaccustomed.

In previous discussions, we had people reporting that the only absolute
pitches they had hardwired would be octaves of c.  For them, \relative {
a' } would be a source of confusion, as would \relative a' { a }.

I think that absolute pitches have enough of an importance in LilyPond
that one is seriously hampered when one is befuddled about them.  But I
don't want to propose making things artificially hard for "educational"
reasons.  LilyPond offers enough complications as it is.

Obviously, the change in semantics will be accompanied with unsolicited
conversion in the case of the old \relative { ... } usage which might
still be found in some files (I think you mentioned that you still have
them), the conversion being to \relative c' { ... } and thus likely to
cover close to 100%, as opposed to the more complex \relative x'' {
... } -> \relative { ... } conversion which needs to recognize and
change the starting pitch in arbitrary ... and thus will only apply to
recognized constructs.

> p.s. Admittedly, I never use \parallelMusic, which I find clutters the
> input and makes true content-presentation separation to be difficult
> (if not impossible). Maybe this is where people encounter a lot of
> their problems with \relative?

Unlikely.  Historically, \parallelMusic was rather hard to use in
connection with \relative.  I think the one thing that worked was to
first split the music in parallel, and _then_ call \relative on the
results.

There were some changes to \parallelMusic which nowadays would make it
feasible to use \relative inside of parallel music.  I doubt, however,
that it is a good idea to do so.  After all, the octave relations and
melody lines are much more likely to make sense in the split results
than the original arrangement.

At any rate, I quite value your opinion as a heavy-duty LilyPond user,
so I am glad to hear that you consider the change something you might
view as a useful tool.

-- 
David Kastrup