Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
vitus
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
42 ms
·
511.
▲
by
vitus
4y ago
This one -- its first occurrence was in '96, several years before Google-the-company was founded. https://www.floc2022.org/history
512.
▲
by
vitus
4y ago
> Is this true? My understanding is that the "crash" is inline with other tech companies losing value in the same period. It spiked because of his antics and then crashed because of his antics. Twitter stock today ($42.52) is b
513.
▲
by
vitus
4y ago
We are indeed still conducting interviews (I had one just on Friday). Whether that transfers to actual offers extended during this time, I have no idea. What I can say we are absolutely not doing at this time is rescinding offers, even for
514.
▲
by
vitus
4y ago
> The only thing that can sometimes bite you here is that str is iterable, if you expect a list of str and you only get a str and suddenly you iterate over the chars. Python isn't even unique in this regard! You can iterate over a s
515.
▲
by
vitus
4y ago
> Nobody would write a make_appender that takes a span argument, because it makes no sense. I don't agree with that. If you can guarantee that the data pointed to by the span will outlive your appender, then it's safe. And if y
516.
▲
by
vitus
4y ago
I feel like a better comparison would be to use std::span (C++20) to mimic Rust's slice. Otherwise you might be tricked into thinking that adding // hey don't pass in a temporary auto make_appender(std::vector&
517.
▲
by
vitus
4y ago
Wouldn't you prefer (committed + 1) / completed? Or (committed + 2) / (completed + 1)? In the limit, you want this to approach 1 (namely, you complete more or less everything you commit). A bad situation is where you're
518.
▲
by
vitus
4y ago
> Why would an AI writing assembly give it an advantage? There are lots of non-human-optimized options that aren't raw assembly. Any intermediate representation, for one (e.g. LLVM IR, C-- used by ghc, Python/JVM bytecode, abst
519.
▲
by
vitus
4y ago
RIP, XMPP + federation (fb messenger, slack, whatsapp, hangouts/gchat, and more used to be accessible through a single client. That said, I recall federation was mostly unsupported by the various hosts, so you still needed to be logged
520.
▲
by
vitus
4y ago
Even easier: open up Chrome DevTools, add a breakpoint at the eval, step into function. Repeat a couple of times. Comment at the top also points to https://freeslots.com/libs/mersenne-twister.js Using those two togethe
521.
▲
by
vitus
4y ago
24fps is still standard for movies, in part because people have associated higher framerate video with amateur home video recordings so it looks cheap. :)
522.
▲
by
vitus
4y ago
> I agree! I just don't think it's what we should be pushing for right now. While I agree that it's not the only problem at hand (and perhaps not the single most useful change the FCC could make), I think any movement is b
523.
▲
by
vitus
4y ago
There are certainly a lot of problems with incumbent ISPs that many other commenters have pointed out: deceptive marketing practices, anti-competitive behavior (ranging from lobby-driven legislation to lock out new entrants, to outright sab
524.
▲
by
vitus
4y ago
5 Mbps down is certainly more than adequate for a single 1080p stream that's been compressed ahead of time. When you have to stitch together multiple real-time streams as in a classroom environment, then the actual numbers can vary sig
525.
▲
by
vitus
4y ago
What if you have a family with four kids, all of whom need to engage in remote learning on occasion? 20Mbps is also entirely noticeable for bulk downloads, whether it's a Steam download or a Windows update. By going from 20 -> 100 M
526.
▲
by
vitus
4y ago
v6 actually has longer headers on average (40B instead of 20B) but is generally more efficient to process because it's a fixed header length. ON the other hand, theoretically IPv4 header length could vary depending on the presence of
527.
▲
by
vitus
4y ago
> allows us to start applying the mathematical techniques of control theory and dynamical systems theory. I'm skeptical that this approach will bear much fruit beyond, say, implementing a basic PID (more realistically IMO, proportio
528.
▲
by
vitus
4y ago
> Consumer PCs don't match this bandwidth because DDR DIMMs generally aren't as fast as LPDDR. What about LPDDR (low-power DDR) allows it to be faster? And, by faster, do you mean lower latency? higher clock rates -> higher
529.
▲
by
vitus
4y ago
We're already close to that point. Low to mid-end AMD cards are already available at / below MSRP (e.g. 6600XT), and nVidia cards no longer command a > 2x premium across the board. At the height of the craze, I saw a five-year-
530.
▲
by
vitus
4y ago
While that's true, the two densest cities in Switzerland are Geneva (~12k / km^2) and Basel (~7k / km^2). NYC sits at about 11k / km^2. Unless 10Gbps internet and up in Switzerland is only available in Geneva, there'
531.
▲
by
vitus
4y ago
Put more concretely: the population of NYC proper (ignoring the broader metro) and Switzerland are pretty comparable (both around 8-9m people), but NYC is about 50x more dense (300 mi² vs 16000).
532.
▲
by
vitus
4y ago
> forgetting to synchronize at all, inconsistent locking At least for these two classes of categories, mutex usage at Google is broadly paired with compiler-enforced annotations to enforce that _all_ accesses to mutex-guarded member vari
533.
▲
by
vitus
4y ago
Or, an even more telling variation: You pick door #1. The host goes and, in order, one by one, opens doors 2-52, conspicuously skips door 53, and then opens 54-100. Would you like to stick with your original choice?
534.
▲
by
vitus
4y ago
> I refuse to believe that (e.g.) the IP address of login.live.com as resolved by OneDrive is always different from same as resolved by Skype. I could believe that, actually. There are reasons you might want to always host different serv
535.
▲
by
vitus
4y ago
On top of that: HVDC scales because resistive loss in power transmission can be reduced quadratically by increasing the voltage (assuming fixed load, so current drops to compensate). (Loss is also increased linearly by transmitting more pow
536.
▲
by
vitus
4y ago
To be fair, this is 0.5% of all processing that Google does, not just encryption / decryption. > it took ~0.7% of Google's processing power to encrypt and decrypt RPCs, along with a corresponding amount of memory. > As of 20
537.
▲
by
vitus
4y ago
Re: illiberalism: https://budapestbeacon.com/full-text-of-viktor-orbans-speech... "in this sense, the new state that we are building is an illiberal state, a non-liberal state. It does not deny foundational values of l
538.
▲
by
vitus
4y ago
> As software capable of zero-click exploit, Pegasus requires no user interaction to operate: ... As a result of a simple click on the URL, the spyware was granted unlimited access to every information stored on the iPhone. That's a
539.
▲
by
vitus
4y ago
S&P 500 growth has historically outpaced inflation by 6-7% annualized over the ~65 years of its existence. That cannot be explained solely due to central bank policy and/or inflation. Naive extrapolation would posit that keeping yo
540.
▲
by
vitus
4y ago
> Buying the dip isn’t some secret strategy. Time is the secret strategy. Ah yes, the good ol' "time in market beats timing the market". Of course, that assumes that markets trend upwards in the (very) long term. Which...
More ›