> But spreadsheets were probably the last quantum leap in the field.
> We're overdue for another.
I'm working on it, and do think I have a solution. Unlike Eve, for example, you can (and we do) create (almost said "write", but there's no textual code) an OS kernel in it.
And it's nothing like the approach Eve is taking, which I consider to be nothing like why spreadsheets were (and are) successful.
To me, Eve is nothing more than textual datalog programming + a high-level stdlib + a UI wrapper.
The only area we agree is that logic programming is an underutilized paradigm, and in my visual programming environment, I do use it for business rules (where it absolutely excels).
I tend to agree that Eve isn't quite the right direction. The REPL is important, so that's good. And one-click deployment, one-click forking of development environment, those things are critically important. But from there...
I agree the high-level stdlib is a problem. I think what we need is an easy-to-dip-your-toes-into environment that provides a path down into "real programming" and ideally down to metal.
I think the big problem isn't exactly UI or language or high enough level APIs, it's the way the APIs are structured. The two big issues I see with most tools are:
1) tendency to layer opaque frameworks on top of opaque frameworks which makes debugging down into the system difficult, if not impossible. (i.e. React) In this sense I think it's interesting that Eve is trying to build on top of itself. Anywhere there's an opaque layer you are putting a massive learning curve for anyone who is grappling with the finer details of that API.
2) declarative interfaces. (i.e. Ember) These require the programmer to have detailed knowledge of the inner workings of the runtime. Feels like we need tools that are made of primitives that the consumers of the tools have a clear path to understanding. There is this idea that the user will never have to understand how the tool works, but my experience is that you always eventually do. You always need to understand the layer below you, so the task is to make those layers easy to dip into... not to try to insulate your users from the details.
I agree with you that APIs are where things fail, and without a good solution for that, you're not going to make much progress. I've spent a ridiculous amount of time working on that problem (and do believe I have a solution).
I'll be sure to post to HN when my visual programming environment is ready to use.
> We're overdue for another.
I'm working on it, and do think I have a solution. Unlike Eve, for example, you can (and we do) create (almost said "write", but there's no textual code) an OS kernel in it.
And it's nothing like the approach Eve is taking, which I consider to be nothing like why spreadsheets were (and are) successful.
To me, Eve is nothing more than textual datalog programming + a high-level stdlib + a UI wrapper.
The only area we agree is that logic programming is an underutilized paradigm, and in my visual programming environment, I do use it for business rules (where it absolutely excels).