On the header parsing
Hi,
just some thoughts on the current header parsing of Mutopia's software
motivated by (not wanting to install java and) the thought that lilypond's
language is so complex that parsing it should be left to lilypond itself.
I noticed that lilypond has (since at least version 2.8.0) a --header
option which could be used to retrieve header fields from the file,
thereby letting lilypond do the parsing for us. Compared to the manual
parsing this of course takes longer, especially since it seems that one
has to compile the piece in order to get the header information, on the
other hand one does not need to care where the header is placed in the
file, whether there are multiple headers or variables are used inside
the headers.
In addition to retrieving header fields the Mutopia software also inserts a
lastupdated field, a footer and a tagline to the header. So lilypond's --header
option is not enough to stop the need of manually parsing the files. But one
could put these fields into a separate header at the end of the files. In this
way they take precendence over fields of the original header while keeping it
intact.
Maybe going a bit further one could strip the (new) files in the git repository
from the (automatically generated) lastupdated, footer and tagline fields and
add a new field like mutopiaId (I hope that the name is self-explanatory). When
preparing the files for the server one adds the lastupdated field, the footer
field with something like
footer=\markup{\concat{"Mutopia-" \lastupdated "-" \mutopiaId}}
and the tagline in a separate header to the end of each file. One advantage of
this would be that if a update/correction of a piece sent via git is ok, it
wouldn't have to be modified at all. For new pieces just the mutopiaId field
needs to be set. Disadvantages would be the fact that there are now two versions
of the files floating around and I'm not sure about legal matters: For a piece
contributed in terms of a CC license then the only hint about this license in
the files in the git repository would be the copyright field. Is that enough to
comply with the license?
I think that it wouldn't be too difficult to implement it, even in shell by
tweaking and extending the current scripts. I haven't tried so, though. A bit
of thought would need to be spent on the best way of how to handle multiple-ly
pieces.
Any more thoughts on this?
Regards,
Felix
RSS Feed