Hey everyone!
I am relatively new to n8n, but since starting to work with it i have had the chance to experiment with it quite a bit, as well as test it based on numerous different aspects.
I have come to appreciate some of its feature - and while overall i think the platform is great, i have a few ideas which can make it even better.
n8n has the ability to test individual nodes and entire workflows in a click of a button, which is, in itself a fantastic debugging tool.
While this might be sufficient for simpler workflows, as they get more complex - so does the need for better debugging.
Currently, i think the debugger lacks in 2 main ways:
First, when executing subworkflows. While they still show as separate runs in the execution log - it is not exactly user friendly, especially when working with a lot of nested workflows.
Further more, if a sub-workflow raised an error, more often than not it is will be a real hassle to figure out the source of the problem.
Secondly, are loops. n8n supports per-node testing, but things get more complex when loops are involved. You are only able to see the final result of the whole loop, rather than each iteration separately - which can be frustrating when needing to debug a more complex workflow.
To overcome those challenges, i suggest a few changes & new features that will be relatively easy to implement, and can make a huge impact.
First, when viewing past executions, it will be really nice if there was a way to easily “look into” a certain subworkflow node - the easiest solution i can think of is when you double tap the node, you enter the execution of that particular run for the subflow.
Second, and perhaps more ambitious, is a more sensitive debugger, perhaps a new button next to “test workflow”.
The new button will open a “debugger” just like in IDEs, with step over step into and so on. Including settings of the sensitivity of each action might be necessary as well (do i want to skip a whole node when click the next step, or maybe just a single part of it?). Breakpoints might also come in handy - if possible.
Again, to reiterate, my intention is not to be dismiss the current product - rather suggest a way for it to be even better.
I will love to hear any further ideas or thoughts regarding this post!
Note: i haven’t worked with the enterprise debugger - apologies if there’s an overlap in ideas.