Study co-author here. We did observe that it's essential to be careful about comparing package version numbers on a per-bistro basis, and there are some tricky

Study co-author here. We did observe that it's essential to be careful about comparing package version numbers on a per-bistro basis, and there are some tricky cases such as the one you pointed out, and rpm epoch numbers as another example. I believe we handled them correctly in the study.