Continuing my thread that hit the bump limit.
Previous Veeky Forums threads
I don't know anything about compression in detail, but if this gives you some compression ideas I'm sure you can use them even without building the whole language
Could it be this language needs some kind of antivirus / anti-malware in the name structure? Like root.security, which might expose functions that scan a local name tree for possible threats.
Other urls found in this thread:
wolframalpha.com
en.wikipedia.org
aws.amazon.com
twitter.com
Alternatively, you could have name subgraphs signed with certificates, but authenticating them is some kind of byzantine generals problem in this case, so it seems like it comes to the same thing. I'm a bit fuzzy on distributed computing stuff, so I'm not sure.
Maybe the entire global name graph could be stored in some kind of blockchain, or market of competing but structurally compatible blockchains?
In fact, maybe the entire distributed store in the back end could be modeled abstractly as some kind of quasi-blockchain like object?
I imagine there's a possible connection here with Clojure's persistent data structures. I don't think this language should have persistent data structures because semantically it doesn't need them, but maybe they can be used in the back end where you have to worry about physical storage and networking
If you have blockchains and cryptocurrency, you don't even need to have an app store for paid software. Since code is data, just call (trade (bitcoin cons structure) (your desired package) (trade safety parameters) (trading venues) [simulated real])
If you hook this language up to a 3D printer, what kinds of things could it make? I don't know much about 3D printers, but the language can probably model them and their materials pretty well, so you could have a very nice link between a cons structure and a physical object along with a precise method for constructing it
Or how about a 3D geometry and texture or surface voxel scanner for input? They have those for computer graphics, I hear.
There might be some kind of cons/decons/uncons/recons quadrality that could help you connect 3D scanners and 3D printers (or just industrial robots and sensors in general), and abstractly model the whole industrial design and manufacturing process. It wouldn't necessarily help you build anything, but the mathematicians would like it
I wonder how hard it would be to take some of the simpler UML type diagrams we have lying around and automatically generate attempted drafts of simple software or at least type hierarchies that implement them, using machine learning on the structure of the diagram itself and the labels on the boxes and arrows