My solution was the development of the nodedev nodes that generate all the files around a node in Node-RED. Each file is a node in a Node-RED flow - again some meta-programming: a flow that generates flow which generates nodes.
What I end up with is a flow which in this case contains all the code for the streaming package. A second flow allows me to publish the package to npmjs and commit to github, all via Node-RED.
Now the advantage is that I don't need to leave Node-RED to extend Node-RED. I don't have worry about remembering all the boilerplate code nor do I have remember the three commands to publish to NPMjs - all that I require is an OTP token.
Disadvantage is that no one understands what I'm doing! But it has sped my development time up enormously and also if I want to update a package, I load the flow from flowhub.org, make my changes, and publish. There is no need to find a directory on my drive where I put the code, I can update my packages from any Node-RED instance and I have a common development environment for all my packages.
Also if a user can use Node-RED, they can then extend Node-RED without using a third-party editor or setup. After all, commonality amongst all Node-RED users is that they can use Node-RED.
A problem that I haven't solved is the resync from github, i.e., if I change the code at github (for example because of a PR) then I need to update my flow - but I can live with that for now.
Just my 2 cents!
