On Fri, Aug 28, 2020 at 9:03 AM Owen Lamb <owendl...@gmail.com> wrote: > > Hi all, > > I've encountered a couple of unexpected regressions for my first commit. > The first one was an issue with tablature stems, which I was able to fix. > The other, however, is baffling to me. > > It appears that musicxml2ly creates slightly different .ly files between > the two commits, for musicxml tests 03b and 71d. In particular, it's > confusing the order of two voices within a staff. The odd thing is that > test 71d appears to have correct output only *after *my commit is applied, > while 03b is correct only *before *my commit is applied (i.e. on the > baseline commit).
I have seen this too. I suspect there is some hash table randomization in the xml2ly code that makes the result unpredictable. It's not related to the fonts, so I'd just ignore it for now. -- Han-Wen Nienhuys - hanw...@gmail.com - http://www.xs4all.nl/~hanwen
