Naive ideas before getting familiar with the code

I’m currently setting up my desktop PC to be able to run Windows and Linux at once with remote desktop.
First step is to download Windows 10 (I don’t have an internet box in my new apartment yet, so it may take a while).
I’ll take that time to introduce you to an habit of mine, and doing it here instead of paper (I’m a pen and paper person).

Whenever I’m involved in a new project, with things never done before, I like to write down a bunch of ideas I could have about it.
You could expect those ideas to be off topic, simplistic or long already thought of. And… yes ! Exactly. But that’s the goal.

My belief is that when you gain a thorough understanding of a project and its constraints, your ability to think creatively about it is more precise and more in line with reality, but as a counterpart it becomes also somehow more constrained.

I play a lot with machine learning, mostly genetic algorithms, evolutionary neural networks.
I more than once found myself stuck, and found solutions reading my early notes, (Talking about my code to people stranger to the project even non-coders may bring great inputs to !).

Now, it’s never direct solutions I find doing that; rather, it may highlight a perspective that my mind would have considered too naive to consider at some point, though it could in fact stem some fruitful thoughts.

What I usually expect to find reading that kind of notes in the future is a mix of stuff obvious, naive, or sometimes plain stupid, but with some bits formulated in a manner that triggers out of the box thinking.
(When you have no clue, you’re out of all boxes ! That’s my attempt at taking advantage of that fact)

(I really feel like writing three more paragraphs to justify how dumb everything I’m going to write may sound. But I hope you got the idea of what this is about at this point :slight_smile: )

Bunch of random ideas in no particular order

  • No forms => Going from UML modeling to code, spiking even forms

  • Machine Learning All data models created by users may constitute datasets for machine learning models to infer and suggest fields or related models

  • Full blown Micro-services printer Totally Hollywood feature: you make a diagram of your micro-services architecture, and write sentences describing what each of them should accomplish. That’s it. From that little input everything is generated

  • Promote Meta With guides, promote and insist on that practice (and I mean, induce an actual shift on what devs day to day work consists in): Whenever a feature is to be implemented on a given project, try and find the most general case. And from that, as a coder, never again just implement a feature, rather write a generator that has desired feature as a use case.

I have others (on paper) that are yet to formulate properly.

1 Like

Hey @ValentinAnybox, no need to write a disclaimer before sharing your ideas :slight_smile: This is the idea behind brainstorming to just share whatever goes through your mind.

Also, I am a strong believer that we can have very powerful intuitive thoughts and ideas that just come to us from nowhere and are actually very relevant ! This is how the most amazing inventions came to life.

You mean : having graphical, drag&drop designers instead of form-based ? So : UML like diagrams for data, but also graphical design for the UI ?

This is a plan to have ! And to go with the self reflection principle of Generative Objects, this idea would be to integrate the capability to model and generate graphical applications. Thus the capacity to model graphically applications on Generative Objects would also be bringing the capacity to integrate graphical design functionalities in all applications generated from Generative Objects.

We need to find open source framework for graphical design on the browser. Maybe to check : what holochain code generator is using for graphical design :
Or maybe interesting pointers here :

And here for open source graphical design of UI :

Yes there are definitively things to do in this area !

Not fully sure I follow what you mean here …

Promote Meta was rather unclear. (And also, spiking forms was likely a typo, but I think you get what I meant).

What I meant is: A dev, having to implement a feature on a context GO would be usable, could implement the feature, or have as a default way to look at the situation: “Can I, instead of implementing my feature, find a more general case that would include that feature and implement in GO” ?