When updating KNIME to 4.5.1 I noticed that the Workflow Executor node, which is part of the “Integrated Deployment” nodes, does not have the 3 points to insert a new dynamic port, for example data.
Was there any usage update or could it be a bug in my installation?
It is interesting that when I download a knime hub workflow with some usage example, the node has a dynamic port, however when I select the same node from my Node repository I cannot include it
Hi @natanaeldgsantos ,
indeed the dynamic ports (three dots) are not there anymore on purpose - the node should figure out the correct ports on its own. This works if you use the output of a “Capture Workflow End” node:
But apparently it does not work if the workflow was gotten from a “Workflow Reader” - this is a Bug I filed under Ticket AP-18532.
Indeed its interesting that the dynamic ports are still there if dragged from the Hub. Do you mean the three dots or the ports itself by that (I’m wondering because your screenshot does not show them)? Can you use this behavior as a workaround for now?
Thanks for reporting!
hah, I was too quick there: this is actually not a bug, but a feature! My apologies @wiswedel for the fuss
In the Workflow Reader node, there is the option to “Remove input and output nodes”
From the node description:
Remove input and output nodes: If checked, all Container Input* and Container Output* nodes are removed and implicitly represented. This allows one to, e.g., combine the resulting workflow segment with other workflow segments via the new implicit input and output ports.
It indexes the workflow and finds the corresponding Container* nodes as they represent the API of the workflows (aka input and output ports). Only when these input and outputs are removed, the ports will be available at the Workflow Executor, i.e. the Container* will be treated as dynamic ports. The concept here is to view the workflow fragments (square black ports) as a workflow with address information where data is in- and output, as usually (i.e. as initially conceived) the workflow fragments come from the Capture Workflow End node.
However: the wording in the configuration of the Workflow Reader is arguably something we can and will improve.
yes, i understand now
I ran some tests and it worked
Hi @LukasS I have not used capture workflow till now but it seems like another (maybe easier) way of “call local workflow”
Is that the case or are there other use cases for that?
in short: yes it is similar and can also be used to call a local workflow.
The capture workflow node pair is intended for integrated deployment (view the extension on the Hub) but is of course not restricted to that.
The Workflow Executor is part of the KNIME Workflow Services Labs Extension which is designed to be more versatile and efficient than Workflow Invocation.
This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.