More of a theoretical question: I am looking into drafting a proposal for a Python-to-Kotlin conversion of a legacy project using the
https://martinfowler.com/bliki/StranglerFigApplication.html strategy, converting units of Python code to Kotlin a bit at a time. We'd be using shared/bridged runtimes, probably over JNI, rather than using jython. My question there is what the "edge of the application" would be considered in that case, for purposes of
IO
. Doing
unsafeRunSync
in every exposed endpoint meant to be invoked from Python seems icky.