Closed romeokienzler closed 3 years ago
Thanks for the suggestion @romeokienzler, is it correct to summarize your idea as:
Implement a preview tab or integrated with the node hover capabilities on Elyra that can display a rendered output of a previously selected notebook cell.
If so, I have some questions:
Note that, until we add support for seamlessly bringing back the resulted artifacts to the user workspace, this would only work if the notebook gets executed on the local user env.
I know this is a challenging and vague requirement, so let me explain a bit why I think it is useful.
In #760 I've requested parameters to be made available to pipeline nodes. The idea stems from usability of no-code/low-code environments like NodeRED, Apache Nifi, etc.
So why not using NodeRED and Nifi instead?
For making Elyras pipeline editor a viable no-code / low-code environment three functions are needed
ETL and Data Mining Tools (incl. Nifi) have a preview functionality where a subset of the data can be shown in a separate tab whereas NodeRED uses a global (global to a particular flow) debug tab where all nodes are writing stuff in
I'm aware that we are talking here about an editor only, so for generating a preview, there is nothing elyra can do. So we leave this up to the encapsulated component (script or notebook).
In case of a script, Elyra checks if there is a function present called "elyra_preview" and the output of that function is renered in a separate tab if present. In case of a nb, Elyra checks if there is a cell present annotated with "preview" and this cell is executed and it's output is presented in a preview tab
Does this make sense? Would it be possible to implement without too much hassle?
@lresende I'll reopen this issue after v2.3.x and make a concrete suggestion on how this could be implemented using Kubeflow
Implement a preview tab which can display an iframe which can be rendered by an optional preview cell the notebook or python script provides. This way, while double-clicking and opening the preview tab on a node the underlying notebook can show sample data