Hi all,
I am interested in the possibility of doing small edits with near realtime visual typeset feedback restricted locally to a few bars centered around the edit location without having to typeset an entire (long and complex) score.
I am aware of the work done by Denemo, yet I'm more interested in the real lilypond typesetting, even when only done locally.
I was thinking such functionality could be implemented by first parsing the entire score (i.e. resolving includes, variable references, traspositions, \mark \default concrete value calculations, etc.) and then either typesetting just the required section directly, or indirectly by first setting a variable to the equivalent ly text input, which, when parsed and typeset, will have produced the exact same fragment of score as the original, but with of the content inline and using only absolute and literal values.
This way, a section spanning several bars or time locations within the score may be typeset in isolation hopefully much quicker than typesetting the entire score. a separate process can then reparse the entire score and set said variable to the new state on any new edit. I am aware of the possibility to render the few last bars or first ones, but not in the middle, or am I missing the obvious?
my hope is that a section of whatever internal representation lilypond uses for representing parsed score (e.g. ast) prior to typesetting can be typeset in isolation. is this possible without a compiler rewrite?
would it actually be able to run quicker? (having to parse the entire score first anyway)
I'm sure there are far better solutions to this problem than my naive solution.
please feel free to suggest any good approach even if requiring some hacking...
Thanks in advance for any line of thought that could put me on the right path :-)
all the best,
Ramon