throwamon

Aren't benchmarks supposed to be more or less reproducible? So just publish the results and the data?

I wasn't sure if you were joking, so I googled "oracle north korea". This is the first result:

https://www.upi.com/Top_News/World-News/2020/10/07/North-Kor...

Sounds like OCaml is exactly what you want. Many of Rust's features are even inspired by it.

> viola

Oh, the irony...

I'm a garbage collector, at least when coding in C.

> most people aren't watching the same thing

I wouldn't say that. YouTube is happy to pick semi-random popular videos from 14 years ago and generic crap and [re]viralize them by shoving them into everyone's recommended feeds, so much so that comments like "the algorithm has brought us together again, see you in 10 years" have become a meme.

So... This might be a good opportunity for me to clear up some of the doubts I've had for a long time. I don't really work in "teams", I'm an amateur and all projects I've worked on have been just myself, for myself. So please bear with me.

Most of the things I do aren't really that important, but there's this one project where I need everything to work perfectly, or else I need to know about issues as soon as possible.

The whole thing depends on input I get from scraping, which basically means the data model can change at any time and there's a myriad errors that can happen. Besides from this input data, I also have to check that my "big-picture" assumptions about the model are correct before the actual program runs. In particular, the sample data already determines what values should be possible for most database tables; these tables don't change during program execution. I have a(n also immutable) config file that determines what can be queried from the database. This means I can prove that my config is/isn't correct (e.g. the program should never try to query for an inexistent value) before the program runs. And it should then be possible to turn these values into types to further tighten up the code and my workflow. But creating types from database values just feels all sorts of wrong. Or is it not?

All I really want is something that is DRY while also 100% ensuring the consistency of data before I run the program. I guess my idea is: Everything that can be automatically proved about the program before it runs, should be, and it should be stopped on its tracks before running if it's provably incorrect. I don't see how Clojure does that.

From everything I've read about Clojure, it sounds like the perfect fit for this kind of thing, but in the end I always conclude I'm too dumb and undisciplined to rely on myself to use it correctly. I make too many mistakes that boil down to assuming something has a different type than it actually has, or that it has a field it actually doesn't, I forget about checking important conditions, etc. And, to be quite honest, I'm too lazy for my own benefit, so I need the system itself to force me to do it the right way, while Clojure assumes you're responsible enough.

So I guess my question is: If I switch to Clojure for a critical project, how can I sleep at night?

Related: "Probably the most complete python dependency database" - https://github.com/DavHau/pypi-deps-db

> This data allows deterministic dependency resolution which is required by mach-nix[1] to generate reproducible python environments.

[1]: https://github.com/DavHau/mach-nix/

See also: Pyret, a Python-inspired functional language by the creators of Racket - https://www.pyret.org/

See also: https://news.ycombinator.com/item?id=31478308

And no, I didn't get the URL wrong ;)

(meaning: just read TFA)

I love what the Internet Archive does and I assume it pisses off many big fish. How do they get away with so much content? What guarantees are there that they're going to be able to preserve digital history in the long term?

nor*

I know I'd pay just to never see an ad again because I don't care about them at all. If one slips through the cracks I do whatever I can to suppress it as soon as possible while actively denying any form of sensorial engagement with it. I suppose this is true for many people and paying might be more convenient than using adblockers, installing alternative apps, using PiHoles...

To anyone looking for a HN-like community with tags, Lobste.rs seems like a pretty good one. It has a few other interesting features, such as an invite tree (so you're more or less responsible for the people you invite) and "hats" ("a formal, verified, way of posting a comment while speaking for a project, organization, or company").

Any plans to implement something like Phonenix LiveView? I've never used Go, but from what I've read I suppose it should work really well given Go's concurrency model. (The name "Livebud" gave some hope as well.)

If you haven't already, check out Nix/Guix.

> Why would the omega 6/3 ratio matter?

You might also ask yourself: Why would the HDL/LDL ratio matter? Why would mineral ratios matter? This pattern is not unique to Omega 3. Search for "micronutrient ratios", also timing the consumption of certain vitamins and minerals, fiber, etc. Nutrition isn't just about putting enough molecules into your body.

You're right, I was rude, sorry.

The issue with psychedelics has nothing to do with addiction. Google "psychedelics set and setting" and "psychedelics psychosis risk factors" for a start.

> I've lived long enough and seen enough synchronicities to know that reality is far more entangled than a cause and effect standpoint would have you believe.

I've no idea what you mean by that.

> This video is the world we live in.

Reality includes komodo dragons, just as it does kindness. You say nature isn't "humanoid" and then ascribe to it qualities like ruthlessness. You're just as biased and delusional as whoever you're criticizing.

I, for one, have the attention span and knowledge organization skills of a Phineas-Gaged goldfish, so it's good to be externally reminded of the existence of good things from time to time.

Oh, and it's DRM-protected, too! We truly live in the future.

Don't quote me on that, but I believe Om was superseded by Om.Next and now Fulcro? So not exactly abandoned, just evolved?

Cool. Not sure how useful it is considering what other commenters said, but is there anything like this for indices or currencies?

Also, typo: Higest -> Highest

Can't write a longer comment now, but check out siuba, or Polars, or dplyr; there are others. (I've posted about siuba recently but I promise I'm not a shill, I'm just allergic to pandas)

I'm glad you mentioned ketamine because I was about to recommend you psychedelics. I can't say I'm cured from depression -- in fact I'm still extremely close to killing myself -- but mushroom trips have been my saviors. Now I just need to start taking advantage of the improved baseline of mood and openness to experience that lingers for a few days (which is unusually short-lived from what I've researched but still useful) to do these things that are "supposed to feel good" and self-sustain for longer.

Just please do some research as this stuff is not risk-free.

> By generating artifact trees at build time, embedding the hash of the tree in produced artifacts, and referencing that hash in each subsequent build step, GitBOM will enable the creation of verifiable and complete artifact trees while requiring no effort from, or changes in, most open source projects

So, Nix? (Except maybe the part where there's no effort/changes needed...)

For further inspiration, this is a pretty good-looking "dplyr for Python": https://github.com/machow/siuba

There's precedent in Ecto for a "magic" operator (`^` if I'm not mistaken), so it wouldn't be a stretch to implement it here as well.

A bit off-topic, but could someone ELI5 what a lattice is in this context?

Wow, genius idea! I really really love it.

It does get a little tiresome to keep starting over, so maybe breaking down each text into subtexts that each get memorized separately would work even better.

I wonder how well this would work for math and other technical texts.

Edit: I can memorize some lines very quickly, so it would also be nice to be able to skip all the intermediate steps and just completely hide the line, then increasingly backtrack if I forget it.

Another minor UX feedback: maybe align the "reveal" and "yes" buttons so I can press the buttons faster and without having to look at them.