WnZ39p0Dgydaz1

9 ms·

31.
by WnZ39p0Dgydaz1 6y ago
It's just a bad list. It includes hosted Kubernetes solutions, application unrelated to Kubernetes (which may run on top, think Postgres), deployment options, built-in metrics, and lots of other random things. It's as if someone g
32.
by WnZ39p0Dgydaz1 6y ago
The main benefit you get from managed Kubernetes solutions is network/security, update- and scaling-related. These things are important, but they don't reduce the daily operating complexity around deployments (applications) on t
33.
by WnZ39p0Dgydaz1 6y ago
Thanks for the insight, that sounds very promising! Right now I'm operating a k8s cluster on cloud VMs, so not managed, but not custom hardware either, myself and like you said it's pretty much a full-time job despite the relative
34.
Ask HN: Has anyone moved from Kubernetes to Nomad?
15 points by WnZ39p0Dgydaz1 6y ago|7 comments
35.
by WnZ39p0Dgydaz1 6y ago
It's pretty obvious this is just another scam. Why is this on the front page?
36.
Ask HN: Why are so few developers with ML background interested in trading?
2 points by WnZ39p0Dgydaz1 6y ago|3 comments
37.
by WnZ39p0Dgydaz1 6y ago
> Programming is a very demanding trade, and I don't think you can do it just for the paycheck, you need to have some degree of passion for it. If someone disagrees with me about this I'd love to hear different takes. Consider
38.
by WnZ39p0Dgydaz1 6y ago
BTC moved has nothing to do with exchange trade volume. When BTC is traded on an exchange, there are no blockchain transactions involved. It's all "internal" to the exchange because the exchanges are centralized. Trading on d
39.
by WnZ39p0Dgydaz1 6y ago
Two things to consider: 1. A large amount of BTC volume is fake volume or wash trading to benefit the exchange marketing. Particularly CoinMarketCap is an extremely misleading and unreliable data source. Their data is totally useless and me
40.
by WnZ39p0Dgydaz1 6y ago
I recently built a project in both Gatsby and NextJS. Or rather, I built it in Gatsby due to the hype and then threw away everything to do it again in NextJS. I would highly recommend going with NextJS. Fundamentally, they are both very sim
41.
by WnZ39p0Dgydaz1 6y ago
Some of it is due to the language and the general JS tooling being bloated and slow. However, in many cases build time is slow because you're doing something that's slow, like calling a REST API. You are not going to generate 10k
42.
by WnZ39p0Dgydaz1 6y ago
Instead of a Markdown file, imagine your data is somewhere in a REST API, or many REST APIs. Gatsby (and next.js, which I vastly prefer) will query these APIs during the BUILD process to generate your static sites - and that can be slow.
43.
by WnZ39p0Dgydaz1 6y ago
Javascript re-invents "code typing" (Typescript) Javascript re-invents "Promises" because callback hell Javascript re-invents "compilers" (babel) Javascript re-invents "build systems" (webpack, etc) J
44.
by WnZ39p0Dgydaz1 6y ago
I don't know. Maybe it's not abandoned and maintained by someone. I was using the word loosely as in "focus has moved to nng" and it makes little sense to use nanomsg for a new project at this point. In my experience, if
45.
by WnZ39p0Dgydaz1 6y ago
zmq is used in a many large-scale production systems. It's definitely safe to use and has been stable for a long time. nng is relatively new. Personally I have not had any issues with it. I'd look through the open Github issues an
46.
by WnZ39p0Dgydaz1 6y ago
My understanding: Pick zmq if you want something safe and battle-tested with good client library support. Pick nng if you want to be at the forefront of new tech or need some of its unique features over zmq ( https://nanomsg.org&#
47.
Ask HN: Why would someone look for a female CEO?
1 points by WnZ39p0Dgydaz1 6y ago|3 comments
48.
by WnZ39p0Dgydaz1 7y ago
Agreed about poor phrasing, it isn't really about "understanding". But we know that humans are robust through millions of years of evolution, and we know that, mostly, their incentives are aligned with other humans. I don&#x2
49.
by WnZ39p0Dgydaz1 7y ago
As a counterexample, look at how sure people and companies were that AI would take off and solve most problems in the 80s. Historically, you can find plenty of examples for both over- and underhyped technologies. It's a complex world,
50.
by WnZ39p0Dgydaz1 7y ago
> A well-run, transparent, centralized entity can easily become an opaque and corrupted entity over time. And the opposite can also happen. Humans are good at quickly adapting to circumstances and fixing "bugs" in organizations
51.
by WnZ39p0Dgydaz1 7y ago
I don't consider myself an expert here, but I believe there are few or no problems that only blockchain technology can solve. Rather, it's about how it solves problems. If you solve a problem with non-blockchain technology, you
52.
by WnZ39p0Dgydaz1 7y ago
I've built a pretty complex and high performance go-based microservice architecture (~12 services) as backend with testify and gomock for testing. Pretty happy with those choices, never had any issues. The backend doesn't use HTTP
53.
Ask HN: How do you manage/schedule your life?
3 points by WnZ39p0Dgydaz1 7y ago|0 comments
54.
by WnZ39p0Dgydaz1 7y ago
There's a difference between research and engineering. Is this system impressive? Definitely! It's a complex engineering effort, and highly tuned to solve a specific problem - beat the Atari benchmark. Does it teach us anything fu
55.
by WnZ39p0Dgydaz1 7y ago
Meh, I'll take the bait. There are no fundamental new ideas in this paper compared to the preceding papers. What they do is to tune hyperparameters (BPTT length, exploration/exploitation tradeoff, and policy parameterization) in a
56.
by WnZ39p0Dgydaz1 7y ago
I agree with you overall, but while it was relatively simple, DQN was never "elegant". The elegant and most important part of DQN was to use a CNN for learning a good image representation - but CNNs are completely unrelated to RL.
57.
by WnZ39p0Dgydaz1 7y ago
This seems more like overfitting a complex system to a specific problem set.
58.
by WnZ39p0Dgydaz1 7y ago
According to the paper this takes around 53418 hours to train (5e10 frames / 260 frames/sec), distributed over 256 machines that's around 200 total hours wall-clock time.
59.
by WnZ39p0Dgydaz1 7y ago
The reason you may not be getting the answers you want is because your question is not particularly precise. It's kind of like asking "which programming books are good?" - Well, it depends on what you want to build. Some peop
60.
by WnZ39p0Dgydaz1 7y ago
As a socially reclusive nerd I have been training for this my whole life. Finally a chance to show off my skills. You're not getting me, COVID-19!

More ›