\version "2.11.14" \paper { ragged-right = ##t }

\new Staff { \partcombine % first time, no rest { s2 g'4 a' } % 1 { r4 g' b' c' } % 2 \partcombine % second time, a rest is displayed { r4 g' b' c' } % 2 { s2 g'4 a' } % 1 }

On a related note, see Issue 913.

A new workaround is available in versions after 2.13.35.
The method of displaying the combined parts can be forced, so that when a rest and skip are combined, we can at least tell Lilypond display the voice with the rest : \new Staff { \partcombine % first time, no rest { s2 g'4 a' } % 1 { \partcombineSoloIIOnce r4 g' b' c' } % 2 } Depending on the situation, the commands to insert before the rest are \partcombineSoloIOnce, \partcombineSoloIIOnce, or \partcombineApartOnce if you want the rest(s) displayed with vertical shift showing which part has the rest.

Thanks for the workaround Keith, it helped on the French Users mailing list today. The issue is opened since more than 4 years now, but it is nice to see workaround coming and shared!

This cannot be fixed until we decide what desired behavior is.

When we \partcombine a printing rest and a spacer rest, what would we expect to see?

I very tentatively propose that we would expect these parts to be considered different-enough to set as partCombineApart, otherwise both parts would have used s4 or both r4.

The change for Issue 4205 (commit 386c1fb71fef99d9ac865ccd1449412c26cef5cc) has changed the output as Keith suggested in comment 4.

An alternative would be to set the rest as if it were shared by both parts. With work on Issue 4233 fresh on my mind, I think that would probably be rather easy. Whether that is more desirable I don't know; one could easily put a rest in both parts to get the same outcome, right?

I believe this original bug report is fixed. With 2.19.16 and forward, we do get the rest printed both time, once in voice one and once in voice two. With the supplied code (s2 versus r4) it does look somewhat ugly, but does appear to be what is wanted.

Verified