\relative does its job just fine on the output of \autochange and \partcombine. The only problem is that \absolute and \partcombine assume they have been given the final pitches, so applying relative afterwards messes up the work of \autochange and \partcombine.
So \autochange and \partcombine defensively prevent application of \relative on their output, which might be less confusing to users that don't realize any relative needs to go inside \autochange and \partcombine These are strange regression tests, in that they test that \relative has been disabled. Maybe we don't want regression tests for this behavior. If you do, can you combine them into one, and write it so that the output looks correct and/or the text describes the correct output? https://codereview.appspot.com/250170043/diff/20001/input/regression/autochange-relative.ly File input/regression/autochange-relative.ly (right): https://codereview.appspot.com/250170043/diff/20001/input/regression/autochange-relative.ly#newcode6 input/regression/autochange-relative.ly:6: in multiple passes, not all of which are aware of the outer @code{\\relative}." What multiple passes ? https://codereview.appspot.com/250170043/diff/20001/input/regression/autochange-relative.ly#newcode9 input/regression/autochange-relative.ly:9: \relative { \autochange { c f b e } } The output looks strange, and will look even stranger after you make \relative non-functional. Can you make an example that looks correct when the behavior is correct? https://codereview.appspot.com/250170043/ _______________________________________________ lilypond-devel mailing list [email protected] https://lists.gnu.org/mailman/listinfo/lilypond-devel
