Re: bar line interface / repat structure considerations

[Marc Hohl]

Am 13.03.2013 12:18, schrieb David Kastrup:
> Marc Hohl <address@hidden> writes:
>
> > Am 13.03.2013 10:27, schrieb David Kastrup:
> >
> > >     One reason I am skeptical about this
> > > particular suggestion is that more often than not (like the typical
> > > multi-stanza pop song with a separate ending), D.S.a.C is _intertwined_
> > > with a repeat volta.
> > LilyPond already allows for nested repeats, but one has to
> > take case for ambiguities, that's right.
> "intertwined" is not just nesting.  It's more goto-like, and
> instructions like "Dal segno", and "al Coda" _are_ gotos in nature.  Not
> every goto-like structure can comfortably transformed into a structured
> form.
Ok, that's probably true. My examples/proposals were not based
on real songs where I'd like to see them.
> > > > b) if the user writes \bar "S.S" without defining it before,
> > > > LilyPond does (define-bar-line "S.S")herself, therefore
> > > > allowing the user to write bar lines "on-the-fly".
> > > Sounds tricky.  I am also not sure that it is a good idea to hide that a
> > > particular bar type just has "best guess" support.
> > Oh, sorry, I meant
> >
> > \bar "S.S"
> >
> > leads to
> >
> > (define-bar-type "S.S" "S.S")
> >
> > A warning that one uses a bar line not properly defined
> > can be raised, of course.
> Could even be an error.
Ok.
> > > Even if the user only acknowledges this by writing
> > > #(define-bar-type "S.S")
> > > and thus signifying "I am ok with all the default choices".
> > What would you propose instead?
> Uh, that's what I propose.  Force the user to at least write
> #(define-bar-type "S.S") before using a bar type that LilyPond _could_
> try deriving a default recipe for.
Then I misunderstood you. Ok then.
> > The current interface ignores unknown bar lines (that's not
> > ideal) but forces the user to define bar lines before using it.
> > We can stick to that behavior and raise warnings like
> >
> > (define-bar-line "myBar" "S.S")
> >
> > "warning: you define a bar line without line breaking decisions and
> > span bars.
> > LilyPond uses "S.S" at the end of line and as span bar, too."
> No, that makes no sense.  You warn about something that the user may
> consider totally fine.  I'd accept
> (define-bar-line "myBar" "S.S")
> without further comment.  One could probably even accept
> (define-bar-line "S.S") without further comment, but that implies a
> strong suggestion of naming bar line types after their default
> appearance.
>
> On the other hand, I would not distinguish between undefined \bar "S.S"
> and \bar "myBar" with regard to which deserves an error.
>
> The error is an undefined bar type.  How LilyPond chooses to meddle on
> (if at all) does not belong in the error message.
Ok.

So we seem to find some consensus here, that's great so far.

>