| > I just don't think Go has a place in new code. This is a completely blanket statement. Not only are there very large open source projects written in go which your company more than likely already uses (kubernetes, docker, prometheus, influxdb to name a few) there are also case studies of very respectable performance in go: 1 million requests/min
http://marcio.io/2015/07/handling-1-million-requests-per-min... 1 million websockets in go:
https://www.freecodecamp.org/news/million-websockets-and-go-... event loops in go: (surpasses redis in their benchmarks)
https://github.com/tidwall/evio I see golang more as a progression for JS or Python developers than Java or JVM based languages, as the concurrency model in golang is closer to to pythons + javascript async. The argument that Java, Kotlin, Erlang developers are less likely to move to Go is probably a valid one, but it is completely baseless to say that "go has no place in new code". The concurrency model in golang with channels makes more sense to someone more familiar with python/javascript who hasn't been exposed to threading or other concurrency models. If you can understand basic coroutines/channels you get the benefits of golangs concurrency (and even parallelism) without much effort, or code complexity, with an order of magnitude more performance that python or javascript for the same feature set. The profiling tools in golang are also first class (pprof, trace, ect) With pprof, its easy to see even traces of different coroutines, heap allocations, cumulative allocations, these are even safe to sample on production running services and this is provided by the standard library. There are even built in tools to profile GC. The argument that tooling and monitoring doesn't exist in golang is simply not true. |