Re: Partial compilation

[Jan-Peter Voigt]

On 04.04.2014 16:32, Urs Liska wrote:
> But lilypond probably could assist a frontend with a map from musical
>> flow to paper-flow. If the GUI knows that on page 3 are measures 27-42
>> and the front end tracks an action, where I modified one slur-shape or
>> one pitch, the front-end could trigger a compilation of measures 27-42
>> and not the whole score.
>> Its just an idea ...
>
> Maybe that map wouldn't even be possible. I _assume_ that an editor
> might get that information by itself through point-and-click. But
> having the information directly would of course ease things.
With delayed expressions - they are used for table of contents - it
should be possible to get page infos related to grobs and there musical
positions. ... and now I have to take care not to start to play
schemedoku again ;)

> What seems very useful to me would be something like this:
>
> - compile a score similarly to using lilypond-book-preamble, i.e. in
> individual files per system.
> - write out something like a LaTeX .aux file with information about
> measures in the systems. Editors can parse this file to know about the
> piece. An editor should then also be able to display these individual
> files like one score (maybe as a continuous page).
> - Enable Lilypond to compile just a given system. This would probably
> work similar to skipTypesetting but with the difference that the
> system is laid out exactly as in the full context. Am I right that -
> once the line breaking has been done - a system is a quite independent
> entity for LilyPond?
>
> That way an editor could provide a continuous view of the score and
> let LilyPond recompile any given system without having to redo
> everything else.
> And it would be sufficiently straightforward to let LilyPond (or the
> editor) protest when the line breaking doesn't work anymore (e.g. the
> recompiled line gets longer and would need different line breaks).
>
> For a "music entry mode" this could even be extended by a compilation
> mode that doesn't care at all about page breaking. Let the lines be
> ragged right and simply add measures until the line is full. This
> should reduce a noticeable part of the compilation time.
Or lilypond shall just encode (ragged) all measures visible in a view to
scroll only horizontal the whole piece.

More on this on monday ;)

Have a nice weekend!