Re: Accidentals: Unwanted naturals

[Alexander Kobel]

David Rogers wrote:
> On Fri, Aug 28, 2009 at 08:03, Kieren
> MacMillan<address@hidden> wrote:
>
> > I can't remember who in this thread first suggested that when you play an
> > instrument you "follow the key signature", but this notion is silly — and
> > ultimately harmful (to music education).
>
> That would be me.

David Rogers wrote:
> When *you* play music, you know that you need to follow the key
> signature in order to play the correct notes. Lilypond, on the other
> hand, *never* follows the key signature - you must always type the
> sharp, flat, or natural that you want the musicians to perform, even
> if you have already specified it in the key signature.

For the impatient reader: I'm on Kieren's side... :-)

I see your point, I guess. But I'm pretty sure most people on this list 
are convinced that LilyPond actually _does_ follow the key signature, 
just not the way this very user wants it to be.

LilyPond is built to typeset music as a professional engraver would do. 
But it's also about the flexibility in the output. We want to be able to 
get MIDI output, part combining, extracts from scores, as well as other, 
more fancy stuff.

Suppose I have to - in my view - disregard the key signature, so I have 
to write "f" to get an f sharp-sounding note in a piece with key 
signature g major.
This means I can't possibly use this music definition for MIDI output, 
without specifying the key signature (which is just a shortcut in the 
printout, basically), but I have to adapt it accordingly.

I know. It's possible, like we have the \relative command - but it means 
an additional and highly non-trivial conversion step, if you consider 
everything which can be done in LilyPond.
To me, it's the same as saying: Well, this quarter note is merely a 
third of this \time 3/4 measure. So I'll better specify it as a 1/3 
note, since I may change the timing (which amounts to transposition in 
the pitch context) of this measure later on, but it has to sound the same.
Which is certainly understandable - it's just not how people tend to think.

Okay, you say, I don't want to bother with remembering those few glyphs 
at the start of a line. Let's just write "f" instead of "fis", since we 
don't see an accidental printed in front of it; we can translate it for 
MIDI somehow, using a \relativeKey command or something. But in essence 
this means, I'm bothering about when an accidental is necessary myself.

Now assume you want the same melody in another printed context. Say you 
want additional cautionary accidentals when writing a tenor voice 
together with the bass in the same staff to spare paper (which, by the 
way, also probably means that the clef changes, so you can't refer to 
"the second staff line" or something), but you also want to be able to 
get a version with one staff per voice, where the cautionaries are 
merely confusing.
Right. You can write a converter. But even with this you were forced to 
always reinterprete and re-evaluate your _own_ input when tweaking the 
score, w.r.t. the printed version you just happen to have lying around.

I'm a very enthusiastic fan of not forcing people to think the way 
machines do, but the other way around.
But this just sounds plainly wrong to me.


Cheers,
Alexander