146 karma · joined July 28, 2024
- Similar graphs
- Medical grade baselines
- His whole channel is dedicated to this kind of stuff
- The original post is good, its just this channel is insanely good
Pitfalls to watch our for? Tons of them. Comparison is very different, modulus is different, .sort is different, object destructuring doesn't map nicely to python, lambda's won't map nicely to python, promises won't map to python. Labelled loops won't map nicely to python.
If your JS snippets are truly simple, just LLM translate and manually check. They're pretty good at the simple stuff.
I'd love to see a tool that could combine all these into one executable. (The lanucher and the target of the launcher)
I disagree. The equivalent would be a thread that's never joined, like a thread with an infinite loop
function blockingSleep(amt) {
let start = new Date().getTime()
while (start + amt > new Date().getTime()) {}
}Last sentence I read of that article
I'm actually not a fan of the "bring your own serializer" design. I think just use one serializer thats fast (like protobuf) and then let other serializers be built on top of that (e.g. JSON). I could be wrong, but thats my hunch of a good system.
The limitations section at the bottom of the readme is excellent. Good level of detail and self-awareness. More projects need that.
Sadly, while theres a lot of stuff I like, ultimately I don't think this is the ROS replacement the world needs. Specifically because of step 1: clone the repo (or add as submodule).
Its a terrible move, IMO, to not have the pub/sub system be a binary executable or shared object file. ROS's fatal flaw (among many flaws IMO) is requiring every library-user to be able to build from source. I'm glad it doesnt dictate a folder structure, but I'm really sad to see this project continue the trend of needing to build from source.
Even though I critiqued the UX, the UI itself is incredibly polished. Which is a a pretty big deal for a framework. Its also pretty rare for me to see that in hackernews UI demos.
In terms of being/not-being a product; really, the most exciting part (for me) would be having (just) the doc-editor as a web component. I'll ask in a separate reply about that since others might want to know.
I'm super impressed. I love the names (text not string) and extra fields (like the multiple choice and tensor). But I'm a bit concerned about the complexity. Don't answer these, I figured it out, but this is how it felt: how do I make an array (numbered list) vs a named list (object)? I didn't realize the intro document itself was, well, a document. I was thinking it was a typical website with a sidebar. Script, whiteboard, simulator (??) I though json was just a hierarchy. Schema, Interface, raw data, json. Its a lot of jargon. Is a document a type of json object? Is there a fundamental difference between whiteboard file and a document file? Is a whiteboard a type of document? Can a document be a field in a document? When editing an item, do I need to fill out the interface, methods, snapshot, etc. Where is the save/submit button? (I known their isn't one, but I kinda expect a "I'm adding a new field, okay I'm done adding it and want to go back to a table-of–contents view") I think the UI needs to answer these because it feels overwhelming). You probably even explain it in text somewhere, but let's be real; IRL people aren't going to read the docs.
Good news is, I think its easy fixes; hide the templates the first time someone is making something (only let them make a doc). Hide the methods, snapshots, etc too. Wait till they've done some stuff. Change the "x" button to "save" even though its not really saving.
The windowing system is really impressive, and I've always loved the detached tables of MacOS's spreadsheets. But at the moment I think it adds to the overwhelming feeling.
Don't let my criticisms weigh you down though! I would've been happy if this had 1/10th of the features.
As a normal human though; when overwhelming daily evidence disagrees with scientific findings, usually its a flaw or VERY deep misunderstanding of the science.
I don't see funding or conflict of interest mentioned, but the author has previous publications about videogames not causing increased violence.
While this is easy to agree with, there is a counterpoint (and a counter-counterpoint to that): EVEN a bicycle could weaken a person (assuming they're going the same distance). And LLM's can act as amplifiers (skilled devs can learn even faster).
The counterpoint can be found in failure, alignment, dependency, and control.
- (failure) We know how bikes fail, but think of Crowdstrike, or the Ford Pinto explosions. People have much less of an intuition for failure possibilities of cars, computers, and especially LLM's.
- (alignment) Faster = more fatigue, long trips = usually healthy, more weight = more effort, etc. A bike replacing running still keeps all those things true. The amounts are different, but the relationships are the same. With a car though, longer trips ≠ healthy, higher speed = 0 additional fatigue (100mph while falling asleep), more weight ≈ unnoticeable. Amplification isn't always aligned, and misalignment isn't always bad (books don't align with what they replaced). But when a replacement detaches a human from their instincts, there's a major risk.
- (dependency) The author mentioned dependency, but its important to point out that if a bike breaks, 99 times out of 100 times it's still possible to walk. Bikes rarely create a dependency. But think of Crowdstrike; there was no backup plan for many 911 services. Cars, computers, and LLM's are often dependency-forming (its good business for them to be)
- (control) Manufacturers have very limited control of a bike. Sure, people can't realistically manufacture their own bike. But once a bike is purchased the manufacturer is still pretty much cut out of the picture. Computers and cars used to be more that way, but now they're increasingly less so.
There's a new wave of Rust based tools. Things like sd instead of sed, and fd instead of find. I think they have an opportunity to fix these things. They're not built in right now, and probably won't be anytime soon. But maybe eventually they will be.
Although, on a different note, I don't think they follow the hive or cathedral design that this author discusses so maybe they will end up with their own lack of cohesion.
Let's say there's lots of columns/fields to sort by. Not quite an arbitrary sort (e.g. can't sort by area when given a width and height column) but its still pretty flexible. Is is possible to get a O(1) pagination while avoiding the rolling-shutter problem and deleted-anchor problem?
Getting around the deleted-anchor problem with ordered-ID's seems like just using ID's as a index. Not a bad trick, but seems like it isn't generalizable for all different kinds of sorts.
> Football (206 posts)
Either hacker news really likes the national forensic league, or these LLM-categories are a bit dubious.
Also hmmm:
> American football (7 posts)
> American_football (6 posts)
Man, the author did not beat around the bush, they're straight up naming names! Props to them
> 3. Unshut Labs may terminate your access to the app at any time and for any reason, without notice.
https://converseapp.notion.site/Terms-and-conditions-004036a...
So ephemeral actually refers to your ability to use their app.
Uh huh, cool, let's look at that app's terms and conditions agreement:
> 3. Unshut Labs may terminate your access to the app at any time and for any reason, without notice.
Correct me if I'm wrong, but I think thats what we in the bussiness call "not decentralized". This whole startup seems like a sham, and 25k being low makes a lot of sense.
Receipts: https://converseapp.notion.site/Terms-and-conditions-004036a...