It seems like there is a deadlock on the socket when using a Supply which doesn't occur when using a plain Channel. The same problem doesn't occur when using the identical server and client code as separate processes. I've noticed the same problem using NativeCall functions that create a socket to a server in the same process.
I've fixed the symptom of the problem by moving the client part of the Supply test into a separate script and executing it, which works fine and achieves the aim of the test by exercising the Supply response body.
This probably points to a deeper problem within rakudo or moarvm, but this is a viable workaround for the time being.
Hi, This fixes #33 and #31
It seems like there is a deadlock on the socket when using a Supply which doesn't occur when using a plain Channel. The same problem doesn't occur when using the identical server and client code as separate processes. I've noticed the same problem using NativeCall functions that create a socket to a server in the same process.
I've fixed the symptom of the problem by moving the client part of the Supply test into a separate script and executing it, which works fine and achieves the aim of the test by exercising the Supply response body.
This probably points to a deeper problem within rakudo or moarvm, but this is a viable workaround for the time being.
Thanks.