The old variable definitions followed the corresponding words in the options.
I see the bug in the old code, so left-to-right did not behave symmetrically to right-to-left. If the actual behavior fits the chosen options with your change, then good enough. https://codereview.appspot.com/42000044/diff/1/lily/axis-group-interface.cc File lily/axis-group-interface.cc (left): https://codereview.appspot.com/42000044/diff/1/lily/axis-group-interface.cc#oldcode739 lily/axis-group-interface.cc:739: || (directive == ly_symbol2scm ("left-to-right-polite"))); The original l2r meant that one of the two "left-to-right-*" options was chosen. Below, it means the objects closer to the left end of a line are placed before those further right, so the items on the left tend to go closer to the staff when there are overlaps. https://codereview.appspot.com/42000044/diff/1/lily/axis-group-interface.cc#oldcode790 lily/axis-group-interface.cc:790: if (x_extent[LEFT] <= last_end[dir] && polite) The old "polite" meant that if we are about to place an object that would overlap one just placed, ask this object wait politely until we first to place all objects that fit without overlap. Placing each object on the first pass was called "greedy". (The old logic here was not quite right, because if we are placing from right to left -- the old 'ltr==false' -- then the test for overlap concerns the x_extent[RIGHT] of the object we are about to place. It should have been x_extent[ltr?LEFT:RIGHT] and the converse below. As it is, the loop is quadratic, but does not hurt too badly because it handles objects in just one after-breaking line.) https://codereview.appspot.com/42000044/ _______________________________________________ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel