The idea is:
I want to be able to assign a trigger dependency to another node. i.e. it’s like the “wait” node but the criteria is “wait for node XYZ to finish”. Or like the “merge” node but without the data joining aspect.
My use case:
I have a complex prompt templating workflow that injects a lot of variables into a prompt and passes that to an LLM. Each of those variables is created with it’s own “sub-branch” within the workflow, and many of the branches have their own intermediate LLM and database operations steps and result in data with very different schemas and numbers of input items, so they can take very different amounts of time to execute. Because of this, I can’t reliably use time as a trigger criteria so the “wait” node is not helpful in it’s current form. I basically want something that works like the merge node, but without the data joining requirement. If you simply added a “passthrough all inputs” option to the merge node so it only acts like a trigger node, this would solve the problem.
I think it would be beneficial to add this because:
Having predictable workflow execution scheduling is an essential core capability of this type of tool. Being able to control flow dependencies separate from data operations seems like a missing piece of an otherwise excellent prototyping and automation tool. This exists at the workflow level via workflow trigger nodes, but should also exist for sub-workflows.
Any resources to support this?
(https://www.reddit.com/r/n8n/comments/1g73zlq/one_big_miss/)
Are you willing to work on this?
Certainly!