On Sat 24 Oct 2026 at 17:34:03 (+0200), David Kastrup wrote:
> Paul Crabbe writes:
> > 24 oct. 2026 15:45:43 David Kastrup:
> 
> >> The propagation rules of \relative are basically:
> >> { ... } the last pitch in the set propagates as the reference outside
> >> << ... >> the last pitch in the set propagates as the reference outside
> >> < ... > the first pitch in the set propagates as the reference outside

> Anybody care for filling in the following sentence?
> 
> Even if we were starting from scratch, the current defaults are for the
> best because ...

If you alter the octavation of one note, you only have to find
one other note to compensate for the alteration—the one after
the altered note or run of notes. It makes alteration and
compensation simple and local.

> If we can find great completions for that sentence, it may save us a lot
> of work.

I wouldn't call it a great completion; it's rather prolix.

> And the reason was that for { ... } the behavior makes sense.  For
> << ... >> it doesn't.  Why should any part in << ... >> get to influence
> any other part?  So it makes sense to all start them with the initial
> \relative reference.  And with regard to the outgoing pitch: there might
> be a weak incentive to the first track for conveying \relative outside
> (the dedicated survivor track), but it may actually just be least
> contrived to bounce out the original reference unchanged.

Possibly you're thinking of << ... >> in its musical sense. I'm
looking at the behaviour of \relative strictly as a textual property
of the source, note after note after note.

If you were reading some source code looking for octavation errors,
you would have to hold a mental marker at the note preceding the
opening << for calculating the octavation of each and every note
that's following an enclosed \\. OTOH, LP, as is, only demands
that we look at the preceding note.

Where it /does/ get a little more complicated is if one nests
\relative's, \fixed's and \absolute's within each other, In that case,
one does have to keep track of the last note before an inclusion
in order to know the octavation of the first note following it. However,
I think those constructions are more unusual.

That said, writing input inside a \relative makes sense for the OP's
temporary polyphony, but I don't think it's wise to use it for
wrapping larger scale structures than voices, where one might want
to move/switch/delete/insert several parts of a score around.

My defence for chords, which seem to be an exception, is their
size and locality: the eye uses pattern recognition to skip from
first note to first note, without having to see the other notes.

Cheers,
David.

Reply via email to