|
|
|
|
|
|
by adsweedler
1929 days ago
|
|
|
Awesome app. Do you plan on using it for anything in particular? Or are you just creating it as a passion project. It's totally cool. *** Learning about https://treenotation.org/ (linking this for other people, not for you, Breck :P), and I like what I see. My first impression was: "tree notation is like lisp, but with python indenting". But then looking at it more and I see it's much more like YAML or CSV or whatever. But then I read: > We no longer need to store our data in error prone CSV, XML, or JSON. Tree Notation gives us a simpler, more powerful encoding for data with lots of new advanced features And I didn't understand! Tree notation seems equivalent to these. Like at a certain level, it's all just data. Now, the major benefit is that you're supposed to think differently about what you're doing when using tree notation. Would love to hear your opinion about this conjecture. *** Not getting into the beautiful and amazing work that's been done WITH tree notation (yet), that's a whole other conversation! |
|
|
It's hard to believe, but I use that thing everyday, haha. To me every problem is a nail solvable with my DSL hammer.
> are you just creating it as a passion project.
I want other people to use it, but it took a while for me to be truly confident that it will work and the underlying math is sound (that 2-d languages are better than 1-d langs). So just in the past week formed a corporation (https://publicdomaincompany.com/) and growing the team and are actually going to try and make a good user experience and help people use these technologies to solve their problems.
> Would love to hear your opinion about this conjecture.
I always think of code now in 2 and 3 dimensions. To me after many years this is just second nature but it's not an obvious thing and not sure I've ever met anyone else that does this. I gave an early talk about it in 2017 (https://www.youtube.com/watch?v=ldVtDlbOUMA) at ForwardJS (there should be an actual recording out there on the web somewhere but I can't find it).
Traditionally all programming languages are 1-D, read by a single read head that moves linearly from start to finish (with some backtracking, but always just along 1 axis in order), building up an AST and then going from there. There's no reason it has to be that way, it's just the way things developed with our 1-D register machines.
Human beings do not process language this way at all. Not even close. If you have a physical newspaper by you, pick it up and pay attention to the way your eyes parse things. You'll likely notice a random access pattern, with your eyes moving constantly across both the x and y axis, and you parsing the semantics by things like position, font size, layout, etc. To me it's so obvious that this is the way computer languages should work. Their shouldn't be lots of physical transformations of the code, using cryptic syntax characters like ( and " and [ and { as hacks to provide instructions to the parser. Code should be written in accordance with a strict grammar, yes, but it should also be written in way pleasing to the human eye and the way human brains work.
We should make human languages that machines can understand instead of making machine languages that humans can understand.
Anyway, I'm rambling on and on, but this is the bigger idea than simply the tree notation implementation, which is really just a subset of a whole new world of possibilities in 2d and 3d languages. (here's another recent one showcasing the new possibilities: https://www.youtube.com/watch?t=145&v=vn2aJA5ANUc&feature=yo... — when I write Tree Code I see spreadsheets and vice versa)
Also, when I play with legos now I see code, and when I write code I see legos. That's a hard thing to communicate and an early tool I wrote called Ohayo shows it off a bit, but need to write a single function that takes a tree program and spits out a lego vis (something like LDraw), and maybe that will help explain the idea https://github.com/treenotation/research/issues/33