Re: Google Summer of Code 2015

[David Kastrup]

David Garfinkle <address@hidden> writes:

> Heyo,
>
> I finished my midterms and am working on my GSOC application. Please let me
> know if I've grasped this right:
>
> If I understand correctly, export involves converting natural lilypond
> scheme data into SXML so we can use Guile to reduce the problem of XML
> conversion. And for import, it's the same way but backwards. In contrast to
> Frescobaldi's lilypond syntax parser, this would convert straight from
> lilypond to xml.
> Finally, this project benefits Lilypond because it facilitates translation
> to and from other music engraving software, which is fantastic. 
>
> Beyond that, I have a few questions: 
>
> What are the advantages of the Guile method over Frescobaldi's enhanced
> syntax-parser?

The structure of LilyPond's internals changes quite slower than its
various input methods get extended, and one can also make input a lot
more convenient by defining music functions of one's own instead of
utilizing the predefined ones.

Many of the example files in the LilyPond Snippet Repository work by
defining such convenient music functions.  Almost none of those will be
properly passed by Frescobaldi's parser.  Almost all of them will yield
music expressions that would be amenable to XML export.

> What part of MusicXML conversion is best contained for a summer's
> project?

Export I think.  In its simplest form, it just converts a music
expression into a list and then to SXML which can be consecutively
dumped without any modification.  That would be a tangible milestone for
further work (it would not yet be MusicXML).  I think that is a clearer
path to develop along than the reverse.

In contrast, import material in MusicXML proper is somewhat sparse to
come by and for any import task one needs to decide on proper LilyPond
constructs.  This is sort of easier to decide when one has already made
decisions for export.

> Is it realistic to believe I can complete both import and export in
> three months?

No.  There are separate frameworks to be written for writing out and
reading in.  It makes more sense to spend any additional time on
fleshing out one direction (it will not have complete coverage by the
end of the project) than to start the next one.

> Who would be my mentor in this case?

Well, setting up the framework to arrive at the input music expressions
would likely be something you'd work out with me.  I'd probably be the
go-to guy for most other places of getting stuck.  So it probably would
make sense if I did the mentoring.

> What do you suggest as a goal for midterm-achievement and how should I
> best split up the task?

Midterm achievement would like be having the SXML of the input and
figuring out how one wants to organize the conversion of LilyPond SXML
into MusicXML SXML.

> Could we Skype?

I don't have any voice chat possibilities set up right now.  IRC would
be a possibility, and we both could try setting up Ekiga.  Skype is
unfree and increasingly freedom-subverting software (they have given up
on free protocols and most definitely on any form of user-verifiable
operation or encryption) so I am not really interested in trying to get
it to run on my computer.

-- 
David Kastrup