100% churn, IMO. Language didn't matter at all. In the end, the failure was procedural.
This lasted well into the next Fedora release. That means... this endured for at least six months. The reporter caught it in beta, but Fedora and 'zlib-ng' marched on. Not saved until Mono/Unity caught up.
This didn't have to be hamfisted, but... t'was. Performance gains are make-believe when consumers don't work. Fedora should have waited for 41 after finding issues on 40.
Distributions should do more to curate the software they distribute. This technical aspect was reported - and ignored - in the early stages during the 'sentiment gauging' phase on their own forum. Not just the 'zlib-ng' repository.
https://fedoraproject.org/wiki/Changes/ZlibNGTransition
https://archlinux.org/todo/zlib-ng-migration/