Re: [PATCH [uploaded to Rietveld]] Automatic lyric extenders

[Alexander Kobel]

Hi all.

On 2016-12-25 13:14, Noeck wrote:
> > As extender events are generated automatically now it probably is a
> > good idea to tell yacc / bison to recognize but ignore "__" tokens.
>
> I don't know at which level the __, \overrides and the automatic
> extenders interact. But would it make sense to use __ now, to force an
> extender e.g. on a single long note?
> __ would probably look nicer than the much longer \override.

I just had the same idea.  That'd essentially mean that "manual mode" 
(\autoExtendersOff) falls back to the old behaviour (+/- collapse-length 
vs. minimum-length).

On the other hand, if __ were to be interpreted both as standard 
extenders in manual mode and as forced extenders in both modes, we'd 
have to re-decide whether forced ones should be subject to 
collapse-length (if they occur on natural melismata).
If yes, lyrics in both modes are compatible, but there can be explicitly 
written extenders that are not printed.  (This could be remedied by 
setting the collapse-length default to 0 in manual extender mode, but 
I'm not sure if there is a sane way to do this.)

Also, if __ forces extenders, it might impact the re-usability of manual 
mode-lyrics for different voices (melisma on some syllable in voice A, 
but not in voice B).  And I'm not sure whether I like different meanings 
of __ in manual and auto mode.  How do we want the following to be 
interpreted?

raw = \lyricmode { Foo bar __ baz! __ }
{ c'8~ 8 8~ 8 2 }
\addlyrics \raw
\addlyrics { \autoExtendersOff \raw }

Would this even be a valid use of \autoExtendersOff?  Should both lyric 
lines be the same or not?

But: I cannot imagine a situation where I would not use automatic 
extenders, so I'm a really bad person to judge the need and requirements 
for a manual mode.

> One drawback might be, that old scores have forced extenders then (if
> not treated by convert-ly - but without convert-ly other things are
> wrong, too).

Old scores will be either mostly okay or severely "harmed":

If extenders have been used everywhere they belong, nothing will change 
- except that some really small ones might not be printed due to the new 
rules for collapse-length (which actually fix the previous behavior of 
minimum-length to match its definition).
If extenders have not been used, they will be added, but only in places 
where they should have been used in the first place (fine), or across 
large regions where the should not appear (divisi lyrics situation; bad).

The latter should be immediately obvious and is easy to fix both 
manually or by convert-ly.  I expect that most scores will be of the 
former type.