I use DuckDB and like it. Since many mentioned GB level json in this post, so they have large amount of data. Been column based, DuckDB uses more RAM as row count grows. It can be an advantage or disadvantage depends whether memory is constrained. Traditional row based DB such as SQLite can deal with large database with less memory.
I feel your pain. I also get "chatgpt/gemini/grok... CONFIRMED blah blah" as if these are ground truth. What is even more sad is it sometimes mixed with "from first principles...".
It’s sad. I just expect better from my friends, but this group chat is also full of friends who are chronically online, so I fear they are just so infatuated with AI as it seems they are constantly considering it at every point of their day.
I have a gut feeling that human as a creature learns better when looking at the information from several different angles, both physically and mentally. Been physically I mean looking at the same concept on screen and on hard copy books, perhaps taking notes and mark relevent sentences with a highlighter. Similarly, seeing a concept on physical book and write some short code snippet is viewing the concept from different mental angles. Though I don't have a proof for that and have yet to find a formal research on this topic.
The pedagogy suggests that you retain more when you also have a spatial element to what you are reading - eg you recall not only what the text was but where exactly on the page you read it, and perhaps also how far through the book it was.
Textbook designers know this and use images, callout boxes and insets with case studies/graphs to break up text on pages so that your brain gets extra context to map 'what' to 'where'.
This is (imo) why infinite scroll and mixed order algorithm feeds are such brainrot (even if you are looking at educational content). You try to recall something you read but it was in an ephemeral location in an always changing stream of content.
This is true: a way to remember things is to construct a “memory palace” in a place you know well, where different pieces of information correspond to different locations inside the building.
The solution explorer from Visual Studio flashes into my mind when I think about the codebases I'm most familiar with, and thinking about the code makes the code file come to mind like it's a big piece of paper and it's all represented physically in some form in my mind. I wonder if the way this happens acts like something of an exploit to get those physical textbook benefits.
> I have a gut feeling that human as a creature learns better when looking at the information from several different angles
Ever write a piece of code, pore over each line, test the hell out of it, and only when you actually submit the PR and see the diff rendered in your review tool that you spot some totally bone-headed thing you've done?
Viewing a familiar concept in a new context gives you new insights with surprising frequency. Context shifts the priors in the mind. That's why, yes, combining reference/LLMs/tutorials and comprehesive pedagogic tours on rails gives you the best perspective on a new domain.
(BTW: it's due to this effect that company offsites and retreats are good investments, not wastes of money.)
Absolutely. Whenever I learn a new thing I'll always approach it from different angles - sources from people with varying skill levels (it can be useful to know how another beginner dealt with something!), video form, written text, reading the theory, seeing the applications, trying it yourself in various forms, different constraints etc. For example, since we're on hn - writing some data structure or algorithm you're learning in a few different languages. Doing all this helps massively with building up to true understanding.
(cooking - when learning a new recipe, I'll always find a few versions from reliable sources to compare, see what they have in common, and try to understand the reasoning behind differences)
I learn the same way but there are different people who learn in different ways. Also some people come with some concepts already understood from past experience or education, it’s easier for to pick things up without needing various angles.
I mean really… he hasn’t finished writing a book that’s narratively trapped but he certainly has been writing and working on a lot more than that one book. He certainly is productive at the very least. Is he satisfying your direct desires and expectations about what he should be working on? No. Does that mean his writing device is unproductive or bad or silly or a waste of time? No.
It uses the Python lib Astral, which uses "equations from Astronomical Algorithms, by Jean Meeus", to find location of the Sun in the sky. That method assuming earth motion without gravitational pull from other planets or the Moon, and has accuracy of 0.01 arc degree. Meeus is a rock star in these kinds of calculations. He also gave a truncted version of VSOP87 that has error less than 1 arc second in finding Sun's location.
The method Astral uses for calculate Moon's location has precision of 1 arc minute. For higher precision, LEA-406 [1] can be used.
Content wise the RedHat docs are great, but navigating the doc has a wired feeling that is hard to describe. Everything is black and white, the page has low information density perhaps because of the line space or paragraph space; the typesetting of command line and configure examples is not clear separated from surrounding text; mouse cannot select text of the command line examples; the page top is distracting because it keeps showing and disappearing as mouse scrolls up and down. Somehow the left navigation pane is also difficult to follow, easy to get lost when trying to find a section.
I hope FreeBSD has longer supporting cycle. Its release has a supporting life of less than one year, if missing the upgrade window, then later upgrade is more difficult than others such as debian stable.
There is a saying that many countries have offshored building hardware, aka manufacturing. Now they are offshoring software building to AI. Perhaps the silicon valley will grow a rust belt.
I tried OpenBSD recently and found it behaves very differently from other OS. The same code works on Linux/FreeBSD/Windows but has poor multi thread performance on OpenBSD, async socket stopped working after sending at high speed for few seconds. I am not saying there is anything wrong in OpenBSD, it is just different.
OpenBSD uses a Giant Lock model (simpler code) instead of the fine grained locking mechanism in Linux. And Linux has a some quirks and hacks to improve performance (instead of doing the slow, but correct thing). One example is the USB Gadget thing.
This is not wrong per se, but they also try to reduce/remove this giant lock as much as possible. If you see entries like "Unlocked socket splicing." in the changelog, then this is one more case where the giant lock is not needed and in which therefore all cores can be used.
Is the code you ran on your OpenBSD available (e.g. on GitHub) for others to test?
Curious what async issue you faced, did you report it? Or ask for help addressing?
I just switched to single thread and didn't try to fix the issue. Single thread is fast enough to me, it has throughput ~ 730 Mbits/s in a OpenBSD 7.8 vm on a 7th gen i7 linux kvm host.
There are few dnsmasq (only?) features that are indispensable to some. Examples: sending query of *.example.com to certain upstream servers, or returning NXDOMAIN for phishing sites, or adding all resolved IPs for *.example.org to an ipset for policy routing. The last one works on FreeBSD as well although BSD does not have ipset. The list of *.example_xyz.com can be huge and it is said recent dnsmasq can handle them efficiently.
reply