Re: \relative proposal: putting absolute pitches anywhere within \relative block using @-sign

[David Kastrup]

address@hidden writes:

> This is a breakaway thread from the one with the subject "Proposed new
> available and recommended behavior of \relative"

I don't see what makes it "breakaway".

> I am *OPPOSED* to the proposal to change \relative syntax, as the
> proposal now stands.  I think it is confusing to new users to have the
> first pitch in a \relative block be absolute and the rest be relative.

Shrug.  Consider it relative to f, then.

Personally, I find
\relative { g'' a b c } -> g'' a'' b'' c'''
\relative { f'' a b c } -> f'' a'' b'' c'''
more intuitive than the behavior you propose:
\relative { g'' a b c } -> g'' a'' b'' c''' plus a warning
\relative { f'' a b c } -> f''' a''' b''' c'''' plus a warning

> But I have another idea.  I'm not sure if people will like it right
> away because it means changing/adding MORE syntax, but I think it will
> be MORE useful and more *intuitive*!

More syntax is rarely more intuitive for writing since one would not
think of it without prompting.  I don't consider it more intuitive for
reading, either.

> Here's the idea.
>
> 1. Define absolute octave syntax with the @-sign (let it be a mnemonic
> for _A_bsolute) to be the syntax for temporarily specifying an
> ABSOLUTE PITCH within a \relative block, such that the next pitch, if
> it doesn't use the @-sign also, is relative to the absolute pitch.

We already have \resetRelativeOctave c'' for resetting the reference
pitch to c''.  One could argue for a shorter name, if one considers it
really important.

> 2. Keep \relative X { ... } working the same way as it is (DON'T make
> convert-ly change it around).

There was no proposal to change the meaning of \relative X.  The changes
from convert-ly were meant to make more use of the new \relative without
X, but not for changing the meaning of \relative X.

> 3. Make \relative { X ... } work such the first pitch after the brace
> is expected to be an absolute pitch syntax with the single equal sign.

Sounds like _another_ syntax not mentioned before.

> If it is not, a warning is printed and the pitch is interpreted as
> relative to c' (the current behavior, except for the warning, right?).

No idea what you mean by "absolute pitch syntax with the single equal
sign", in particular as you want to be interpreted relative to c' as
"fallback" which does not appear to make any sense (have a fallback
behave differently than what the check should be for?).

> Why a new syntax?  I frequently find that if I jump to the end of a
> big, long \relative { ... }, then frequently I don't remember which
> octave I'm in.

Welcome to \relative.

> Octave check is not a solution, because if I guess the part that comes
> before the = sign wrong, I'll keep getting warnings until I fix it.

Welcome to \resetRelativeOctave.  Alternatively, just start a new
\relative block inside of the existing one.  Its pitches will be
independent from the outside ones, and the starting pitch is specified
absolutely.

> What is wanted is a way to temporarily jump into absolute note entry
> mode.

Welcome to \transpose c c { ... }.  Which we might call \absolute if
people want that often enough.

> An @-sign comes immediately after the note name, and is followed by
> any apostrophes or commas as necessary to specify the absolute octave.
>
> Examples:
>
> 1.    { c4 c' c@'' c@, }

That looks a lot like "Not invented here" syndrome to me.  You reject
the existing mechanisms as well as proposals fitting with the current
syntax in order to propose ones that purportedly do the same job without
fitting in the existing framework.

> 3.    \relative { c@'4 c g' c }
>
>       This is the same as { c'4 c' g' c'' } in absolute mode.  
>       This would be the form that new users of LilyPond would be
>       encouraged to use in the documentation.

And what would be c'@'4 ?  And how does it combine with octave checks?

> 4.    \relative { c4 c g' c }
>       
>       This is the same as 
>               \relative c' { c4 c g' c }
>       
>       EXCEPT that a warning will be printed about encountering 
>               \relative { X ... } 
>       where X is not specified absolutely (with at-sign).

If you don't want the behavior used, just prohibit it and use convert-ly
to get rid of existing usage.  That's issue 3231 currently on countdown.

> What do people think?

Not a winner to me.  I would not want to spend a potentially useful
extension character like @ on something that does not appear to offer
anything new.

Now you don't like the current proposal, and that's duly noted.
Brainstorming is always exciting and makes one fond of one's own flashes
of genius.  But I think that if you view this proposal realistically,
you'll find that it has several badly defined aspects, it is not
self-explanatory (of course neither is \relative { f'' }, but at least
without any previous exposure to relative mode starting pitches, a
result of f'' is a more likely guess to me than f'''), and it clutters
the writing of pitches additionally.

It is also leads to inconsistencies: why write \relative c' { ... }
instead of \relative c@' { ... }?  The argument to \relative is in
absolute pitch!  Why indeed write \relative at all if the absence of @
is supposed to be sufficient for indicating relative pitch on its own?

-- 
David Kastrup