Right now reading from err and retv files is not a blocking operation, hence clients need to poll retv to find out when the remote process terminates. I think it would be better if retv worked as a pipe, blocking on reads till something it written on the other side or the pipe is closed. Once the read unlocks, clients can also inspect the err file to checkout if an error occurred.
Right now reading from
errandretvfiles is not a blocking operation, hence clients need to pollretvto find out when the remote process terminates. I think it would be better if retv worked as a pipe, blocking on reads till something it written on the other side or the pipe is closed. Once the read unlocks, clients can also inspect theerrfile to checkout if an error occurred.