[linux-audio-user] Rosegarden vs. Note Edit

[Chris Cannam]

On Monday 16 Jun 2003 12:28 pm, Joerg Anders wrote:
> In other projects the MIDI related way of thinking dominates.
> The score editor is regarded as an appendix.

Assuming you're referring at least partly to Rosegarden here, then 
it's not true to describe the score editor as an appendix.  It was 
designed in from the start and was the first editing mode we 
implemented.  As I had previously worked on Rosegarden-2.1 (whose 
score editor is much more like NoteEdit: it only does score, and 
knows nothing about MIDI) the exciting thing for me with Rosegarden-4 
was to set out to do a score editor that was powerful enough to be 
genuinely useful for score, without compromising the sequencing 
capabilities of the application at all.  Of course that's an ongoing 
challenge and I won't claim that we've succeeded yet, but I think 
it's an interesting and worthwhile goal.

> The internal
> data structures must represent the score and the midi data
> must be derived from these data. Why: The mapping from
> score to midi is unique mapping.

Nonsense!  A performance (in MIDI or anything else) contains any 
amount of information not representable in score, just as the score 
can contain any amount of information not representable in MIDI.  
There is no unique mapping in either direction.

The fact is that any program that aims to be a score editor and a 
performance or composition tool at the same time, rather than 
concentrating on one particular area, is going to have to work harder 
and make compromises.  Rosegarden is a sequencer and compositional 
tool, and so (as we agree) will inevitably find some things harder to 
implement in notation than NoteEdit, which is primarily a score 
editor and so is not particularly useful as a sequencer.

(Rosegarden's internal data structures actually represent neither the 
MIDI nor the score exactly, although producing MIDI from them is 
certainly easier than producing score.)

> And another aspect: From the very beginning it was quite
> clear to me: The main problem is grouping: The are
> vertical groups (chords) and horizontal groups
> (beams, slurs, ties, tuplets). Every note can be a member of
> any of these groups, even simultaniously.

Yes, of course.  And basing an internal representation directly on 
that will lead to difficulties in performance or MIDI editing -- but 
I'm not about to claim that that makes it "wrong" to do so, or a 
"basic error".  The two programs are simply different.


Chris