I recently posted a query [1] to the grpc mailing list to clarify a behavior in the Go gRPC server - a panic() in a request handler function would terminate the server process. Now, I know that's not the case in the net/http server implementation - a demo here [2].
However, the interesting thing that I read in one of the replies is that this isn't considered something that is desirable in the net/http package either. but is a historical choice that we wish could have been different. I was wondering if someone here would know about that and why it may not be desirable to have automatic recovery in the http server implementation because even with the automatic recovery in place: 1. The application author/operator user will get to know when a panic has occurred unless the ErrAbortHandler value is used to panic. 2. The application author can still implement a middleware and not let the stdlib's panic handling mechanism get invoked at all to run any custom cleanup for example. [1] https://groups.google.com/g/grpc-io/c/J8T43gWMwnI [2] https://gist.github.com/amitsaha/ec7ce96e47bbfca0e7a52d5bff53485b -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/golang-nuts/CANODV3nGDhPGzAPNyr6KZAuD6uU6jr%3DOYasCjZ%3DJeJbYRiir3A%40mail.gmail.com.
