Common questions#

Pipeline execution#

Messages fail to propagate

Q: The first message from the source is the only one that propagates in the pipeline, any subsequent message is ignored. Why?

A: This often happens because the source node that generates the data messages does not set the version in the messages it sends. As a result, the very first message will trigger a version change from -1 (the default data version recorded in an empty buffer) to None (the default message version set in an empty message). This version change will be detected by all the node waiting for updates on that buffer, but any following messages will have their version set to None, so nothing will be triggered.

As a general rule, when designing a node, always treat the version of any sent messages as a counter of how many messages that node produced. This can be done with a counter local to the node, or by updating the version of the received message:

def update(self, message: Message):
    received_version = message.version

    # an empty message has its version set to None
    message_to_send = Message(creator=self.name)

    ...

    # option 1: use the received version and increment it
    message_to_send.version = received_version + 1

    # option 2: count the number of sent messages locally
    message_to_send.version = self._local_counter
    self._local_counter += 1

    self.transmit(message_to_send)
Plugin nodes are not loaded

Q: I want to use a plugin node that requires the import of an extra module from within the node folder itself, but it does not work. Why?

A: By design, the location of a node also dictates where the node class can look for extra local modules to import. To give you an example, If a node class MyAwesomeNode wants to use a separate module called my_awesome_utils, it will need to know the path where that module can be imported. So, assuming the node code lives in the folder ./plugins/nodes/proc/my_awesome_node, then the node will do something like this to import the extra:

from plugins.nodes.proc.my_awesome_node import my_awesome_utils

Because of this reason, all plugin nodes downloadable from the Juturna repository are expected to be placed in a folder ./plugin within your project folder.

Plugin not found from hub

Q: I want to use a plugin node that is available in the Juturna hub, but once I try to download it the hub says it does not exist. Why?

A: When you download a plugin node from the hub, make sure to use the correct name. Usually node names are in the form of <node_type>/_<node_name>, which represents the actual path of the node in the Juturna hub folder. For example, if you want to download the node my_awesome_node from the hub, you should use the following command:

jt.hub.download_node('proc/_my_awesome_node', destination_folder='./plugins')
On this page