I think constraint solvers are underutilized. Probsbly because people are not aware of them or don’t think in terms of solving problmes based on specifying constraints.
Could you give an example of how these are used? Do you mean things like "IF there's been [4] login attempts in the past [10] mins THEN deny this login attempt"?
Yep, but production rules engines also work in asynchronous cases because they can keep a database of partially matched rules that get updated as new facts come in.
In your case the rules engine would keep track of the login attempts and automatically expire the records once they are old enough that they won’t trigger any rules.
It boggles my mind that nobody has tried making a JavaScript framework based on them; they would be particularly good for the cases where people get burned in React, such as when there are several pieces of information which go into a computation that updates the UI. Rather than having to implement updates for all the cases the compiler figures out all the ways in which a change could force a UI update and makes it happen.
NFTs and crypto ledgers. Deadset someone gimme a web browser where the pages are just ledger entries of assets that are NFTs. Gimme dat decentralised multi hosting of said website assets and decentralised hosting of the ledger I can check them with.
Legit. Art for NFTs. Meh. Securing data transfers w peer hosted ledgers and code as unchangeable nft assets. Yes pls.
What happens when someone hacks your wallet for the NFT but you still hold the physical asset (with your right to do so supported by the US legal system)?
Non-template-based code generation. So many tools depend on interpolation and concatenating strings together resulting in terrible-looking and error-prone code. What you want is an API for building and manipulating a concrete syntax tree-like structure. Bonus points if the nodes have type information, which would make things like codemods even more powerful.
HHAST and LibCST are notable examples in the right direction.
Sqlite for data science. You can do so much with a few lines of SQL, where people usually write error prone, slow python scripts. Importing CSV or TSV data is trivial.
I can imagine a notebook like environment where the results of each step are put into a temp table to use in subsequent steps. Why do you think people are not using it enough? Because they don’t know SQL? Or because of tooling?
Well because usually you pull some data from a service API as csv or Json and you’re not directly pulling from a SQL connection since that isn’t exposed.
Even internally across different teams.
With all the other technologies falling back to json, modern web dev only thinks about SQL as a store for CRUD operations, and not as an integral part of the stack. At least that’s how it seems to me.
It makes only if you're doing a few operations and all of them are on tables. As soon as people need to mix up table operations and other code (custom functions and calculations, REST API calls, linear algebra etc) in the same script, it makes a lot more sense to use a dataframe and have all of your code in the same language.
If there's a way in which sqlite is underrated by data scientists, it's as a storage format - it's a lot more reliable than passing around csv files.
Years ago I wanted to use video game worlds to make TV shows… where you could script the characters and focus on the dialog instead of animating each frame.
Personally I'm interested in bringing videogame-like interfaces to the web so GFX.
Not so interested in GPGPU, which is something I have more experience with. If you aren't guaranteed to have the latest and greatest hardware, GPGPU isn't always worth the trouble.
Barriers are:
• skills (software, artistic, ...)
• Android, Intel GPU and other bad hardware
My third side project in the queue might involved WebGL but I gotta finish one and make progress on #2 before I start on #3.
25 comments
[ 3.9 ms ] story [ 61.3 ms ] threadIn your case the rules engine would keep track of the login attempts and automatically expire the records once they are old enough that they won’t trigger any rules.
It boggles my mind that nobody has tried making a JavaScript framework based on them; they would be particularly good for the cases where people get burned in React, such as when there are several pieces of information which go into a computation that updates the UI. Rather than having to implement updates for all the cases the compiler figures out all the ways in which a change could force a UI update and makes it happen.
Legit. Art for NFTs. Meh. Securing data transfers w peer hosted ledgers and code as unchangeable nft assets. Yes pls.
HHAST and LibCST are notable examples in the right direction.
Even internally across different teams.
With all the other technologies falling back to json, modern web dev only thinks about SQL as a store for CRUD operations, and not as an integral part of the stack. At least that’s how it seems to me.
If there's a way in which sqlite is underrated by data scientists, it's as a storage format - it's a lot more reliable than passing around csv files.
At the uni where I work there is a lot of work on social robots, many of these could be prototyped with no moving parts, only sensors.
https://www.youtube.com/watch?v=5wmsyPsny3s
Not so interested in GPGPU, which is something I have more experience with. If you aren't guaranteed to have the latest and greatest hardware, GPGPU isn't always worth the trouble.
Barriers are:
• skills (software, artistic, ...)
• Android, Intel GPU and other bad hardware
My third side project in the queue might involved WebGL but I gotta finish one and make progress on #2 before I start on #3.