bsima

Templating engines start out with just variable substitution, but inevitably you want to build logic into your HTML, so it adds if/else statements, then before you know it you are embedding your 'real' programming language into the HTML template itself. Think ERB templates with arbitrary ruby code inside them.

It's more robust to go the other way: start with the full-powered programming language, and emit the static HTML data after doing the complex logic stuff.

This looks fantastic! I love htmx, and I'ved used it with Lucid and Clay in Haskell to generate HTML apps entirely server side. I'd really like to use Ludic at some point as well.

The requirement of @override on the render method is a bit unwieldy, have you considered using pedantic's BaseModel instead of TypedDict? I think it would allow you to get rid of @override? Not sure it'd be worth it for the extra dep though.

Something I always look for in a web framework is the ability to run the app from the repl. Is this possible with Ludic? I remember in Clojure, the app was just a function of a hashmap request -> hashmap response, so testing was really easy, you just do something like (app my-request) and inspect the return value, 'my-request' can be a hashmap constructed by hand or generated with a tool like core.spec. Much more convenient than running uvicorn just for a quick test! I recommend adding this use case to Ludic if it's not already there.

as she should

"Coronal mass ejection" would be such a great band name

No. This is a terrible idea

yay, likelihood that we are martians just went up a few basis points

I’m literally working on it right now, dm on X if you wanna pair or something

> I feel I'm profiting off of "the AI hype train"

> While I'm not making any money off my projects

????

It could be both: Fermented meat from scavenging. Sounds gross but actually many cultures still have fermented meat products in their cuisine

As tome mentioned we don’t quantize, all activations are FP16

And here are some independent benchmarks https://artificialanalysis.ai/models/llama-2-chat-70b

Why wait? This is pretty much what Groq has in hardware, just need the software layer to do the review process.

If you can so easily and cheaply generate a new version of the app/widget, why would you even use traditional version control?

Prison is it's own kind of League of Legends

I was excited to see Haskell in one of the screenshots (“difference detection”) only to realize that Haskell is not in the list of supported languages :(

Check out compute blade https://docs.computeblade.com/

Hey this is pretty great, and the code is so simple. I guess it only works if you have the sources checked out somewhere, which isn't the case for all build tooling and package managers, but I could see an extended version of this that hooks into the standard package managers to fetch the required information to complete the report.

If you can also hook into an accounting system (eg plaintextaccounting.org) then you could also calculate the whole dollar amounts to donate as some percentage of income from the product.

"Only a Sith deals in absolutes"

Why is a new and cheaper desalination invention in the news every 3-4 years?

Yes and what’s more is that natural selection is a loss minimization function, similar to gradient descent

Why is this shocking? Evolution itself is a learning system

Hasn’t this been the case for years? Or did something change with Usenet/Google groups recently?

Remote work (because I need to be near family), then salary/total comp. Those are the only two things I really care about. I take advantage of insurance and 401k match etc when I can but it’s not the deciding factor.

I mean, given the options, I prefer palatable foods to unpalatable foods.

> I did an inventory of my recent feature flags and realized that about 80% of them aren’t there to roll things out to specific populations, or do any sort of A/B testing, but to hide unfinished code.

Maybe it’s just me but having unfinished, dead, or scratch code in a production codebase really annoys me. Either finish your work or delete the unneeded code. More than a few times I’ve sunk time out of my day into investigating some code path only to realize it’s completely unused.

How can students know what the teacher tells them is true?

My Tesla Model Y works great. As an "appliance" I am super happy with it.

Most other smart appliances I've used, however, have not created joy in my life, and I try to buy the oldest and dumbest appliances I can. For our new oven last year, we specifically looked for something with real knobs and an LED display, zero smart features. (Viking is the last company that makes these afaict.)

Turns out software is hard, if you're gonna make a smart appliance you have to build an entire software company. Tesla managed to do that but not other companies have.

Definitely usable for those cases. You might have to do some extra work to build the necessary libraries though.

It’s closer to scheme I think, being a lisp-1, even though it’s originally implemented in Common Lisp.

I love Shen and highly recommend the Book of Shen and Tarver's other book on Logic and Computation.

The crazy thing about Shen is that its type system is sequent calculus, which means when you define your types, you are literally writing the same language that mathematicians use when they prove things about type theories.

Why would you use it? Because you can leverage a crazy amount of power by writing your own type theories in your programs.

Is it practical in today's corporate world? Probably not, but there is at least one case study in the real world https://www.youtube.com/watch?v=lMcRBdSdO_U

> I picked up Lisp during my Freshman year ... I do hate the corporate world (and have trouble staying anywhere long) with a passion and would very much like to find a way out.

maybe we should work together

a daughter