Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
zozbot123
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
9 ms
·
31.
▲
by
zozbot123
8y ago
BTC transaction volume is going to be kept fairly stable due to the existence of a dynamic transaction reward. If the cost per transaction goes down, this might imply that demand for "higher-value" uses of the BTC network is indee
32.
▲
by
zozbot123
8y ago
Rust does not solve all race conditions, unfortunately. Even world domination has its limits! It does solve simple data races , which means that use of concurrency within Safe Rust is indeed memory safe. This is better than what some
33.
▲
by
zozbot123
8y ago
The data is so slow in modern-day terms that it wouldn't make a dent in their mobile-data cap anyway. We're talking 9600 baud, or 9.6kbps, for data over the mobile voice channel. Now, SSH'ing (or RDP'ing, for that matter
34.
▲
by
zozbot123
8y ago
It's not just 9/8 more RAM, but 9/8 more memory bandwidth on the bus. So, yes, a pretty nasty performance hog for typical use.
35.
▲
by
zozbot123
8y ago
> an embarrassingly large amount of the software I use on a daily basis for productivity runs in a single thread. My expectation was that UI blocking would be rare- it isn't, particularly with Chrome, Firefox, and Slack. Jira in the
36.
▲
by
zozbot123
8y ago
It's notable that Krugman is now saying that there was "lots of room" to lower interest rates during the 2008 recession - he certainly didn't see it that way at the time, and was very vocal about this view. Does this mea
37.
▲
by
zozbot123
8y ago
Why does a SemVer spec have to deal with that? Why not have a "gestalt" number/indicator for purely informal use, but keep the SemVer part as is?
38.
▲
by
zozbot123
8y ago
> Leaks Reveal Random Musings From Some Guy In Army-SOF Who Doesn't Even Know What The IMF And World Bank Are About, But Somehow Thinks They Might Be Used As "Unconventional Weapons". FTFY.
39.
▲
by
zozbot123
8y ago
NEON is the CPU SIMD feature, it has nothing to do with the vc4 GPU.
40.
▲
by
zozbot123
8y ago
The "good part" of Rust compared to its closest alternatives is the machine model, which is straight C. You can even directly transpile C into Unsafe Rust.
41.
▲
by
zozbot123
8y ago
> Similarly, some pretty nifty stuff in libraries is possible due to the flexibility of these languages. You pay for that "flexibility" quite directly in lack of verifiability and of adequate performance. It's not a free l
42.
▲
by
zozbot123
8y ago
> Or you might have a load of methods that actually throw 'Not Implemented' errors because you are forced to match a massive interface but it only has to do one thing ... The other thing is that you're sometimes forced to
43.
▲
by
zozbot123
8y ago
The Raspberry Pi provides documentation for their GPU architecture, so it would be possible to provide support for that within open source machine learning frameworks. It would involve quite a bit of work, though, and the RPi is not real
44.
▲
by
zozbot123
8y ago
Rust has extensive metaprogramming features as well, so hopefully we'll be able to build a comparable framework starting from that.
45.
▲
by
zozbot123
8y ago
> "a Lie group is a group object in the category of manifolds" The point is of course to motivate the definition of a Lie group. Whether such strong motivation counts as "a better understanding of Lie groups" is la
46.
▲
by
zozbot123
8y ago
Somewhat less of a "textbook", but good at showing applications of category theory - Fong, Spivak: Seven Sketches in Compositionality https://arxiv.org/abs/1803.05316 (Notably, this includes application to
47.
▲
by
zozbot123
8y ago
The point of the category-theoretical definition is to explain how these "rules for how they interact" might arise, and in what sense these might be understood as 'minimal' rules. These are not "equivalent" de
48.
▲
by
zozbot123
8y ago
On the contrary, the significance has everything to do with the category-theoretical pattern, albeit the reasoning for that is not immediately obvious and has to do with a further category-theoretical tool known as the Kleisli category.
49.
▲
by
zozbot123
8y ago
The problem with "a faster approach that lets [you] get to applications" is that it will depend a lot on what applications you have in mind. And often it's not even faster in any real sense - there's a whole lot of po
50.
▲
by
zozbot123
8y ago
TL;DR: the Youtube AI overweights heavy viewers (people who have more "engagement" with YT) and thereby ends up naturally promoting the Youtube equivalent of clickbait. YT creators notice, and make their videos more and more like
51.
▲
by
zozbot123
8y ago
Look, it's 2019. "Setting up category-theoretic language" is something that should begin in the first lesson of the first course in formal, non-K12 math, starting with the use of ETCS or a comparable "structural"
52.
▲
by
zozbot123
8y ago
> because C++ templates and Java-style generics are not able to operate at their level of abstraction, people that are only used to thinking in these terms fail to map them into their "comfort zone" and usually give up. That&#x
53.
▲
by
zozbot123
8y ago
And "these axioms" are indeed the axioms of a monoid operation (for multiplication), plus some sensible notion of "consistency" with the underlying abelian-group structure. Which is all that the "monoid object"
54.
▲
by
zozbot123
8y ago
Many graphics cards/chipsets/etc. do accelerate 2D operations, unfortunately there's no feasible unified model of a "2D hardware accelerator" the way there is with 3D, so there's no unified support for it in
55.
▲
by
zozbot123
8y ago
So far, Wayland is removing legacy cruft and replacing it with newfangled cruft. The Wayland protocol itself is not even well specified in the way that X is - the whole thing is very far from true maturity. Just like, e.g. Pulseaudio, whi
56.
▲
by
zozbot123
8y ago
It's also a decent target for C, if you use cc65 (though proper reentrancy is indeed going to be a challenge). But it's hard to describe C, FORTH or PASCAL as the most "natural" high-level languages for the platform, gi
57.
▲
by
zozbot123
8y ago
> If someone quickly pulls an usb drive, plugs it in another system, and then plugs it back in to the original system, then flushing writes could cause massive data corruption That's user error, though. The kernel should react to r
58.
▲
by
zozbot123
8y ago
> which is to stress the rule you can do the same thing to both sides of an equation (worrying about things like squaring both sides or multiplying by zero later). It's a pity that they're being so ambiguous here, because expla
59.
▲
by
zozbot123
8y ago
Essentially all CISC ISA's were designed for "assembly being written by humans". But I'd still choose RISC-V as a first platform to learn assembly for, before anything else. 6502 is particularly crappy in that it only
60.
▲
by
zozbot123
8y ago
Rust has hygienic macros which one can use to build EDSL's. It's not that hard.
More ›