Am 01.10.2020 um 12:24 schrieb Jonas Hahnfeld via Discussions on LilyPond development:
So I think it's more likely that the issue got exposed by commit 5a957021a3 which runs lilypond once per document passed to lilypond-book instead of per included file. This increases the chance of being unlucky...
Another topic, but possibly related: Could this potentially affect regtests, that use (ly:set-option 'warning-as-error) or similar? In the sense, that the state change triggered by ly:set-option leaks through to other snippets?
Btw., is it still good practise to use the (ly:set-option 'warning-as-error) (ly:expect-warning ...) construct in regtests? I vaguely remember a discussion about that some time ago on -devel...) Cheers, Michael
