Could you not have used a simple thread pool model in your case? That would be a trivial parallelization of your network requests with limiting on the resources (number of parallel connections)
Otherwise when the logic gets complex I try to model it as independent threads each having a well-defined role (e.g. exclusive write access to SQLite) communicating with channels. Although I tend to code these things in Rust, the language who taught me that philosophy is Go.
Otherwise when the logic gets complex I try to model it as independent threads each having a well-defined role (e.g. exclusive write access to SQLite) communicating with channels. Although I tend to code these things in Rust, the language who taught me that philosophy is Go.