Doc: NR - Add 'known issue' for \partcombine (section 1.5.2 Multiple voices)
Probably here: http://lilypond.org/doc/v2.21/Documentation/notation-big-page.html#automatic-part-combining
On Dec 26, 2020, at 13:29, Thomas Morley wrote:
Hi, the following code gives bad output.
%% Start
\version "2.20.0"
one = { e''2 \tag #'score f'' \tag #'part fis'' g''1 }
two = { e''2 d'' g'1 }
\removeWithTag #'score \partCombine \one \two
%% end
it looks like tags are not respected for the length of
skip
, but both are counted, thus the length is off.
Dan Eble replies:
I would call this a known issue, and there are other features that similarly interfere with \partCombine. If it ever seemed to work in the past, it depended on the specifics of the music.
\partCombine is a two-pass feature. In your example, the first pass does not know that you will be removing tagged music, so the information it records is inconsistent with the music when it comes time for the second pass. If you want to use \partCombine with filtering, unfolded repeats, or other transformations, you must transform first so that the music is in its final form before the first pass, something like this:
\partCombine { \removeWithTag … \one } { \removeWithTag … \two }
Thomas follows up with:
the initial example works at least since 2.8.0 (yes, I've a working 2.8.0) A long time... Tbh, I couldn't figure why it works before 2.19.21, though, if it doesn't anymore, it should be documented either as it is done for \relative already or in "Known issues and warnings" probably together with \relative.
See also this post on the user-list: https://lists.gnu.org/archive/html/lilypond-user/2020-11/msg00243.html