Hi all, probably what I am writing now is not new to most of you. About a year ago there was a discussion regarding the license of Lilypond, triggered by Urs' question about the future of OLL. Again and again the documentation was referred to, which says that Lilypond is a compiler that translates the source code into a PDF. For God's sake, I don't want to discuss the licensing consequences again, but I want to point out that this representation is not exactly complete. In fact, each source file is translated into a Lilypond internal executable, the execution of which then generates the PDF. This architecture is representable in XML, if it is possible at all, only with extensions to MEI or MusicXML. So the goal should only be to implement the graphical representation accordingly. But this also means that structures that serve a better organization of the lilypond source code will most likely be lost during export and re-import.
Several solutions for the export have now been mentioned. Behind these are three concepts, all of which have their justification: 1. convert source-based (python-ly accessible through frescobaldi). 2. lilypond internal generation of an intermediate code 3. scheme based generation of an intermediate code If the sources comply, i.e. do not contain Scheme, then the python-ly solution is quite charming, fast and batchable. But I would find the internal generation of an intermediate code, as envisioned and developed by Jacques, the nicest. The Scheme based solution I started from the pragmatic consideration that if it works, it can be quickly adapted and deployed in different environments. I'm very glad to see this discussion revisited. Maybe something more can be developed together in this direction. For this reason, I am also pushing the discussion to the devel list. ;-) I'll be on the road for the next week, but I plan to get back to the topic after that. Cheers, Jan-Peter Am 07.10.21 um 22:51 schrieb Thomas Morley: > Am Do., 7. Okt. 2021 um 13:32 Uhr schrieb Jean Abou Samra > <j...@abou-samra.fr>: >> >> Harm, >> >> Le 07/10/2021 à 11:46, Thomas Morley a écrit : >>> Not sure Jan-Peter's approach is the best method ... >> >> What makes you think so? >> >> Best, >> Jean > > Maybe my wording was misleading. > > I tested ly->musicxml with > (1) openlilylib, i.e. Jan-Peter > (2) python-ly > (3) Frescobaldi > (4) https://github.com/de-wolff/lilypond.git > > Then tried reimporting the resulting xml-file via > (a) musicxml2ly > (b) xml2ly > > All results were terrible. Here I stopped frustrated. > I did not look into any code, thus I simply don't know which one is > the most promising approach. > > Cheers, > Harm >
