Looking to make some code as multi-core friendly as possible via frequent use of go-routines. However, I'm new with golang and missing something apparently basic.
As way of example, I'd like to spawn a go-routine that creates and holds open an rpc communciation channel to an altcoin wallet. Skipping a bunch of irrelavent stuff, the code looks something like:
package wallet
var client btcprcclient.Client
func Server(c chan int) {
client, err := bitrpcclient.New()
defer client.Shutdown()
c <- 1 // signal main routine client is establish
<- c // wait for the main routine to signal us to shutdown
}
func Status() {
blockCount, err := client.GetBlockCount()
}
package main
func main() {
c := make(chan int)
go wallet.Server(c)
<- c // wait for the wallet to be ready
wallet.Status()
c <-1 // terminate the wallet connection
}
If I append the code in the Status() function into the Server() function it works fine. However as pseudo-written, I get an invalid memory address panic when Status() references the global-to-wallet client variable.
I guessing the "right" way of doing this is to rewrite Server to act on channel input, calling a local status() function (and/or anything else I wanted to do with the wallet), but my mind is confused on why the current way doesn't work. Even if main() & Status() run on core 0, and Server runs on core 1, the box still has a common memory model, so the var client's address should be the same???
