Re: Comparing LilyPond with Sibelius, Finale, Musescore etc

[Carl Peterson]

On Tue, Jul 23, 2013 at 12:44 PM, Richard Shann <address@hidden> wrote:

On Tue, 2013-07-23 at 17:23 +0200, pls wrote:

>  (both MusicXML and their corresponding PDF/PNG files) as reference files: http://www.musicxml.com/music-in-musicxml/example-set/.  They cover quite a broad spectrum of music notation.  I would simply open these files with different applications and save the rendered scores each time as PDF files without changing / tweaking anything.  Then we can compare the resulting PDF files.
>
> Of course this doesn't necessarily tell anything about the quality of music engraving of the compared applications.  It rather shows the quality of the file format conversion of these programs.

Yes, for this reason I suggest we do *not* do this, as it will distract
attention from the main point that people do not understand, namely that
just by inputting the music they want to play into LilyPond they can get
a nicely playable score; whereas if they input the music into a
drawing-based program they will have to position things by eye, using
the mouse.
(There is a secondary point, that if they alter the music in a LilyPond
score the re-positioning of everything else takes place automatically,
which often it will not with a drawing program).

We will not help people by replacing this insight with observations
about how bad musicxml2ly or, worse still Denemo's musicxml import is.
Well, in fact they are not so bad, inasmuch as it would be
self-defeating to import all manner of typesetting information into
Denemo or LilyPond, these importers are there to save typing in reams of
notes and durations basically. But, we will not communicate the main
message this way.

So what we need is some musicXML files which just contain some basic
information, e.g.

notes durations and markings

the sort of thing someone might expect to type/click in to a program to
tell it about the music they want.

This would take some donkey work, though (potentially stripping out
information about beaming, slur positioning ...), and it *may* not be
needed. A first stab might be simply exporting scores from the
commercial programs in musicXML and then reading them back. I did this
with MuseScore http://denemo.org/compare#Example_2 and the gives a good
insight into how much hand-tweaking is needed in MuseScore. This would
not illustrate the point if Musescore exported more information to
musicXML and imported more back and it may not work for other programs
which may do this, but it *may* work just fine.

This may be what you're getting at with the musicXML idea, but what about doing what we usually do to demonstrate lilypond...take a reference score, and set it up with no manual edits? So, for example, in Finale you would be able to connect slurs from notehead to notehead, but not adjust the curve in any way. In LP, you would add the parentheses and nothing else. This eliminates any issue of musicXML translation and trying to get the musicXML figured out may end up being like the post a few weeks ago where the poster decided it was easier to re-input the score than to deal with converting software.

 

Richard

Carl