Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
perlgeek
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
11 ms
·
181.
▲
by
perlgeek
1y ago
> The actual bottlenecks were, and still are, code reviews, knowledge transfer through mentoring and pairing, testing, debugging, and the human overhead of coordination and communication. I can relate :-) Our team maintains a configurati
182.
▲
by
perlgeek
1y ago
I'm not super familiar with this, but I think many of the gene therapy advances did involve some form of machine learning to find relevant genes. Some might also use protein folding simulations, and the best ways to do that involve nar
183.
▲
by
perlgeek
1y ago
... and then there was this Phase-1 trial in Japan about re-growing missing teeth. We're really making astonishing progress in many different areas in medicine these days. They are mostly pretty narrow, but also so awesome that ~20 yea
184.
▲
by
perlgeek
1y ago
I really like my lists to be deterministic. Seriously, I've read things about lists and nondeterminism a few times in this thread, and I can't help but wonder if "you guys" (functional programming nerds, maybe?) use the
185.
▲
by
perlgeek
1y ago
There's nothing embarrassing about not knowing the details of a sub-field you're not deeply involved in. Like, why would you embarrassed about not understanding the details of MPLS networking if you're not a network engineer
186.
▲
by
perlgeek
1y ago
nothing, IMHO.
187.
▲
by
perlgeek
1y ago
> Would you consider those people obsessive in the pursuit of traveling? Could be, depending on how much they give up for traveling. > The reason why I'm asking is that my impression seems to be that there is a lot more acceptanc
188.
▲
by
perlgeek
1y ago
> Is there anything particularly "neuro divergent" about having a hobby? No, but how obsessive the pursuit of that hobby is, that's the question. There are some model train enthusiasts that, over their life time, spend sev
189.
▲
by
perlgeek
1y ago
I'd say it's very hard for a powerful nation to not suppress somebody in the long run. Just think of any powerful nation (or group of people, or whatever), and try to think of somebody they have oppressed, or are still oppressing.
190.
▲
by
perlgeek
1y ago
What is the evidence for that? If Israel wanted them to leave, wouldn't they seek cooperation with a nation that is willing to have them, and organize mass transports there? At least I haven't heard of any such thing.
191.
▲
by
perlgeek
1y ago
From Israel's perspective, Palestinians are a problem. Long term, they have a few options: 1) Give them their own state. This is difficult for quite many reasons, and Israel (by which I mean the current government) doesn't want th
192.
▲
by
perlgeek
1y ago
Hugged to death, use https://archive.is/1JT24 instead
193.
▲
by
perlgeek
1y ago
It would be really cool to have an array of space-based telescopes spaced out evenly in the Earth's orbit around the sun, and use each as relay for the others that cannot directly communicate with Earth, because the path is blocked by
194.
▲
by
perlgeek
1y ago
For me, the visual proofs of simple sums (like The sum of the first n odd natural numbers is n²) works pretty well for me. For the more geometry-based ones where you have move triangles around and so, it's often not obvious to me that
195.
▲
by
perlgeek
1y ago
Y'know what, I'd have no problem paying for my "core services" if it were that easy. What I have a problem with is paying for potentially so many services: * phone * email * whatsapp, because others use it * signal, beca
196.
▲
by
perlgeek
1y ago
Our models use single atoms (or single molecules), and for those we have pretty good models that we can solve numerically, at least. In a gas, the atoms or molecules only interact weakly, so you just get some known effects like a line broad
197.
▲
by
perlgeek
1y ago
Stealing session tokens is a common technique to get around MFA, deployed by malware everywhere. I don't like it, but frequent reauth does limit the blast radius of stolen session tokens.
198.
▲
by
perlgeek
1y ago
A use case is when a sub or method needs to do different things based on the data type of the argument. Example: turning a data structure into a JSON string, solved with multi dispatch: https://github.com/moritz/json&#x
199.
▲
by
perlgeek
1y ago
Probably because typing in Python is optional and kinda bolted-on after the fact, and you really need a good type system to get the most out of multimethods. In particular, type declarations, as they were first introduced in python, did not
200.
▲
by
perlgeek
1y ago
It also looks familiar to multi dispatch in Raku, formerly Perl 6. See for example https://docs.raku.org/language/functions#Multi-dispatch
201.
▲
by
perlgeek
1y ago
I know several people with ADHD who have several songs stuck in their had at the same time. If they try to engage with a different tune, they now have two in their head. And it goes on, up to six and sometimes even more. Tough luck.
202.
▲
by
perlgeek
1y ago
The answer to "why?" is really just "Evolution". It's not satisfying at all, but for example sleep could have been an evolutionary accident that happened to have enough upsides and little enough downsides that it pe
203.
▲
by
perlgeek
1y ago
The common wisdom is that that girls often speak earlier than boys.
204.
▲
by
perlgeek
1y ago
As somebody who tried to do a PHD in optical communications, this is 100% correct. I wonder if meta material might provide such nonlinearities in the future.
205.
▲
by
perlgeek
1y ago
> A bit like when people tell you to learn Morse code, not to learn it letter by letter. Fun fact, during WW2 there were lots of encrypted transmissions over Morse code, and lots of folk (often women, in the UK at least) had the job to t
206.
▲
by
perlgeek
1y ago
Doesn't practicing with random notes become very boring? I imagine it would be far more engaging (but also far more complicated) to tap into an archive of songs and present those randomly, either selected by or transposed into the key
207.
▲
by
perlgeek
1y ago
In my head, I just substitute "AI" with "machine learning" or "statistics". > and I feel the AI boom is driven by hypothesis that general models eventually outperform specialized ones given enough size/
208.
▲
by
perlgeek
1y ago
> Where are the AI-driven breakthroughs? The only thing that seems to live up to the hype is AlphaFold, which predicts protein folding based on amino acid sequences, and of which people say that it actually makes their work significantly
209.
▲
by
perlgeek
1y ago
Stronger (tensile and compressive strength) doesn't necessarily mean harder. Maybe hardened steel tools can actually cut it fine.
210.
▲
by
perlgeek
1y ago
You can see both from the videos and the photos that the sails weren't set, so I can't have been wind-powered at the time of the accident. Operating a large, wind-powered vessel in a harbor or near shore is very tricky and dangero
More ›