I'm still thinking a lot about what a "point literal" or a "vec literal" means in the context of visual programming. In text we have "string literal", it has a bunch of affordances. In visual/spatial/tangible/etc programming, we'd probably want other primitive types with their own set of affordances. So point and vec, obviously. I wonder what else we'd want, and how it would look and act.

grammar of graphics makes me think about data spaces. a dataset occupies a place within a data space, which is not a million miles from a table (dataset) SQL schema (space). You can look at a dasaset as a tabular dataframe OR slice into a million possible configuration of a data chart. So dataframe or chart are projections of the same underlying dataset. I quite like things that allow you to view the same thing from different angles. Your examples make me think you are thinking lower level but I do think its more user friendly to think at the higher level.

I'm thinking of "point literal" as, like, a little ball I can hold in my hand, and "vec literal" as an arrow between two balls I can juggle.

ah right, useful marks in a space. Then I would say vectors/arrows have a start and end, but its also useful to have a infinite line (a ruler?) for binary space partitioning. Is a circle a thing you consider in scope or is that a composite? its a point + radius, could be defined by just an arrow though

One example I remember from several years ago was 'bouncy numbers', which are these oscillating points. In general I'm a fan of the proliferation of primitive visual/dynamic types because even though you can represent them in text they are really their own thing. https://mobile.x.com/prerationalist/status/1440442305710723072

Sidebar: Is there a good way to view content on X without logging in?

Some content (like this one) may be on Internet Archive: https://web.archive.org/web/20210922105958/https://twitter.com/prerationalist/status/1440442305710723072

Ah that's great, thanks. Really neat idea.

For pre-X tweets, there's also nitter.net, e.g. https://nitter.net/prerationalist/status/1440442305710723072

Ah that's even better. Thank you!

https://github.com/akovantsev/wayback-url browser ext

The "what does literal even mean" question is why I'm so obsessed with this. As for precise editing, yes.. maybe. Sometimes. It depends, right? Compare with numbers (floats, esp), rather than strings, in this regard.

wiki's definition boils down to "*literal* is a textual representation", huh. in that sense – vpl vec literal is just a ui widget, regardless of what it looks like.

Text is a UI.

as soon as you transition from text file to "graphical canvas" – "correct" choice of default widget becomes less critical, as you already require more interaction from user, so basically you can have widget with 10 tabs where each is different projection with different amount of controls. And you are no longer constrained with "what user has on the keyboard to fast/conveniently replace current literal text fragment with"

So interesting but I have no clue what you’re hinting at. Do you expect textual representation or “graphical language”?

Florian — not sure who your question is addressed to.

Phew you went into full detail in the podcast. I now better understand what string literal means. Would the equivalent in visual programming be 1) A circle representing a drag handle 2) A literal circle, as in, graphical shape drawn to the end user?

A side note / observation: electronics schematics represent behaviour that is very different from what synchronous, sequential programming languages represent. Any 5yo, any CEO at a whiteboard, any EE, etc. knows that a closed figure represents an asynchronous unit and that schematics represent inherent, massive parallelism. Programmers tend to think in terms of clockwork, synchronous, sequential, blocking functions which is the opposite of what is drawn on electronics schematics.

@Florian Schulz I'm experimenting with 1. The tangible things I'm programming with form the "source code" of my program. So while I could add some way to display them to the end user directly, for now I'm just using them to measure and calculate and then (indirectly) perform drawing operations. Like, the visual program is abstract, edited separately from the world of UI. (This is fun for me because historically I've been all "there should be no distinction between program definition and runtime environment")

@Ivan Reese I have been working on this idea for a while: What if I use a diagram from Category theory to display and manipulate source code. Reading source code and writing source code through the diagram. I have been following a little bit the discussions here regarding visual programming but not extensively. The point is that when you describe a category you need to define the composition method, ie how two arrows compose in the diagram and what are the objects (points in the diagram). It is all based on the fact that category theory can describe any abstraction. On top of that it allows you to have some formal ways to move between different categories or between levels of abstraction.

@Theo Gpl check out algebraic visualization design for sure, it's an interesting twist on 'compiling to a diagram' https://www.researchgate.net/profile/Gordon-Kindlmann/publication/265470662_An_Algebraic_Process_for_Visualization_Design/links/541030e10cf2df04e75b7707/An-Algebraic-Process-for-Visualization-Design.pdf c.f. an explicitly semiotic take https://core.ac.uk/download/pdf/9991583.pdf tom larkworthy's latest might also resonate https://observablehq.com/@tomlarkworthy/cell-map

@Jasmine Otto is think the confusers and the hallucinators were illuminating there. I was just calling them monoids over source code or the diagram, optimisations or transformations.