A solution for tedious tasks of node related icons

[esnosy]

Right now we have node icons in many places:
1 - Node shelf
2 - Node headers
and soon node add menu

we should find a way so that we define icons once and it is referenced in all of them, or at least twice, one in Python, and once in C
also for the toolshelf we should try to automate some pieces of code because right now the code is very duplicated, there should be some sort of auto generation

this is a discussion for near future

[ReinerBforartists]

Code design pattern is an interesting topic for sure.

I am no friend of long defines lists. I am more a friend to implement the functionality where it is used. If technical possible of course :)

I am already unhappy how the header icons in the nodes are implemented, by the long case list. But i am also pragmatic. It works, and i don't know of a better way, so i leave it alone.

But imho such external dependencies are not good. I prefer another design pattern. I just know the german word for it. It's called Selbstverwaltender Code. Google translates it with self-managing code. I think that's the OOP programming approach. Everything required should be available where it is required, and not cluttered across several external classes and files. When we have a property with an icon, then the icon should be defined in the property. And not in a long list elsewhere. That's for the add menu for example. We should add the icons directly in the operator property if possible. Saves also lots of searching.

In the end everything working is allowed. As told, i am highly pragmatic in this regards. I am green with everything that works.

[esnosy]

Interesting indeed, I agree with you 100%, We should look for a good design pattern for this issue, but if it works we could leave it until a better way is found

[esnosy]

I also agree the long switch case is not the best, it works until we find a better way

[esnosy]

For the record this idea finally became true with #2308