Re: Reasons why a LilyPond-to-MEI conversion should be developed

[David Kastrup]

Simon Albrecht <address@hidden> writes:

> On 24.10.2015 03:18, Paul Morris wrote:
>> LilyPond’s internal scheme data structure is a good target for
>> import and export
>
> Or even more so the music stream internally used by LilyPond after
> Erik Sandberg’s work. IIUC there is some doubt as to how complete the
> features he describes in his master thesis actually are, but if they
> were this would be very promising.

They are not in LilyPond.  There is no tangible or recognizable thing
like a "music stream": stream events are delivered to separate contexts
separately.  There are a few functions for writing out stream events in
a rereadable form: those have mostly stopped working some time over the
years due to bitrot.

One can register listeners like the event recorder used in autochange or
the part combiner does.  The results are so-so since not everything is
done via stream events and iterators do a considerable part of the work
without going through stream events.  Also some events are delivered to
specific listeners that would just not be around when replaying a "music
stream".

> To quote what I already wrote to the mei2ly-application list:
>
> “Erik Sandberg divided LilyPond into two separate modules, one to
> parse the input file and create the music stream, and another to read
> the music stream and layout the music (This change happened somewhat
> after 2.6.0, IIUC).

But not _in_ 2.6.0.

> His master’s thesis, to be found at
> <http://lilypond.org/pdf/thesis-erik-sandberg>, describes this
> model. Notably, in chapter 7.3 he describes that he has implemented a
> facility (where to be found?) for separately running each of the
> modules.  Appendix D contains examples of music stream code, which is
> in the Scheme language.  And chapter 9 sounds very promising in terms
> of our current discussion.”
>
> Do we have a possibility of contacting Erik and getting his opinion?

Sure, you can.  I'm not sure whether he even still has the respective
code (I seem to remember that he had some sort of catastrophic crash
without backups but may be misremembering or misattributing).  But
whether or not he has, chances are rather slim that parts of the code
could be usefully adapted to work with LilyPond these days.

--  
David Kastrup