Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
Dunedan
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
24 ms
·
241.
▲
by
Dunedan
5y ago
Fun fact: If you download a file containing colons with popular web browsers, they'll try to be helpful and will replace the colons to avoid problems storing the file on filesystems which don't support colons. Took me a while to f
242.
▲
by
Dunedan
5y ago
Can you elaborate on this? Parsing ISO 8601 formatted dates doesn't result in naive Python datetimes: >>> datetime.fromisoformat("2021-10-24T20:17:41+02:00") datetime.datetime(2021, 10, 24, 20, 17, 41, tzin
243.
▲
by
Dunedan
5y ago
For finding bugs caused by unexpected inputs I also find property based testing very useful. For Python there is the excellent hypothesis library for doing that: https://hypothesis.readthedocs.io/en/latest/
244.
▲
by
Dunedan
5y ago
Aren't larger displays the reason for the need of bigger batteries in the first place?
245.
▲
by
Dunedan
5y ago
There's another approach which works surprisingly well: If you use Nginx and PHP FPM you can utilize Nginx FastCGI cache [1] to cache dynamic contents. That even allows serving a stale version of an asset while fetching an up-to-date v
246.
▲
by
Dunedan
5y ago
Luc van Donkersgoed wrote a blog post about inconsistent AWS APIs just a week ago [1]. Funny coincidence. [1]: https://www.lastweekinaws.com/blog/how-aws-dumps-the-mental-...
247.
▲
by
Dunedan
5y ago
Ignoring SMT (as others already mentioned it's not used for AWS Lambda), I also expect different vCPU scaling and they also hint at that in the detailed blog post for this feature [1]: > Workloads using multithreading and multiproce
248.
▲
by
Dunedan
5y ago
> There is no way to print a fully generated CloudFormation template. I'm no fan of Serverless Framework either, but accessing the generated CloudFormation template which will be used to deploy the AWS resources is actually pretty s
249.
▲
by
Dunedan
5y ago
What is the maximum memory size you tested the AWS Lambda functions with? From your article it seems you did try with 1024MB. Did you try with more memory as well? In case you don't know, you need at least 1769MB of memory to get a ful
250.
▲
by
Dunedan
5y ago
> (Berlin's pirate party once won a surprisingly high share in the state parliament but wasn't able to fill all seats, because they were polling way lower when compiling their list.) That's not correct. They had 15 candida
251.
▲
by
Dunedan
5y ago
> > "Racked up a several-hundred-thousand dollar bill in a couple of hours." > This is enough to rent big server from Hetzner / OVH for like forever and have person looking after it with plenty of money left. That&#x
252.
▲
by
Dunedan
5y ago
This doesn't work for phone numbers if you want to do something programmatically with them. In Germany it's for example common to include the international prefix _and_ a trunk prefix like this in written phone numbers: +49
253.
▲
by
Dunedan
5y ago
> It's not like being just a little more flexible with input formatting is a major feature with huge implementation costs and security risks, either. Just in case somebody considers implementing phone number parsing and normalizing:
254.
▲
by
Dunedan
5y ago
> Because no one knows how to build a sensible web service any more, the page loads a 1.3MB JSON file containing every number that Barclays has. While other comments have already mentioned that it's actually "only" ~190kB
255.
▲
by
Dunedan
5y ago
TSMC offers both DUV (N7, N7P) and EUV (N7+) for 7nm [1]. [1]: https://en.wikichip.org/wiki/7_nm_lithography_process#TSMC
256.
▲
by
Dunedan
5y ago
> […] AI has a massive value to every aspect of the company's work. That's also just wrong. During the recent "Tesla AI Day", when asked during Q/A, Elon Musk specifically mentioned that they intentionally use ma
257.
▲
by
Dunedan
5y ago
> The T1 had zero impact whatsoever on getting Linux running. If you define running as "Linux boots", than this is correct, but as the the T1 chip provides access to the Touch Bar which is necessary to have function keys, I
258.
▲
by
Dunedan
5y ago
> It's pretty amusing that they assembled it with a boring old construction crane. This crane is neither boring, nor old. It's an engineering marvel itself. It's a Liebherr LR 11350 [1], one of the most powerful mobile cra
259.
▲
by
Dunedan
5y ago
That's what Netflix does: https://www.businessinsider.com/netflix-encourages-employees... (version without paywall: https://archive.is/mdMr9 ). Also check out their culture document, which is quite inte
260.
▲
by
Dunedan
5y ago
> Buy up ASML's inventory and next N months' production at an exorbitant premium? Which inventory? As far as I know ASML is already churning out as many EUV machines as they can and yet they are still fully booked for the next
261.
▲
by
Dunedan
5y ago
> TLS 1.2 is the minimum supported protocol, as recommended by RFC 7525, PCI DSS, and others In addition to offer encrypted connections, each publicly available SMTP server must accept unencrypted connections according to RFC 2487 as wel
262.
▲
by
Dunedan
5y ago
I agree, but I'm confident Noctua will tackle that with the next iteration of their 140mm fans which is currently scheduled to be released in Q1 2022 [1]. [1]: https://noctua.at/en/product-roadmap
263.
▲
by
Dunedan
5y ago
Not that this matters much, as the chance to get an IP address delisted is pretty slim anyway. I've completely given up to try to get my personal mail server delisted, as I can't even get Microsoft to tell me why they blacklisted
264.
▲
by
Dunedan
5y ago
> There is no need to spend extra for 140. A quality 120 is negligibly higher CFM then a quality 140 at the same RPM. Where do you get these numbers from? Take Noctua fans for example, which are arguably some of the best on the market. A
265.
▲
by
Dunedan
5y ago
> The obvious logical choice is for Tesla to start retrofitting Superchargers with CCS now that it's open and eventually ween Tesla owners into using adapters that Tesla already makes. European superchargers already offer CCS ports.
266.
▲
by
Dunedan
5y ago
> If you're not large enough for that to apply or don't plan for huge growth, personally, I'd strongly reconsider AWS. Amazon's APIs and infrastructure automation can give small teams a lot of leverage to run massive
267.
▲
by
Dunedan
5y ago
These 125ms are only the startup time of the MVM and don't include additional latency introduced by optimizing the code package and the involvement of the placement service. You can also avoid the cold start penalties entirely, if you&
268.
▲
by
Dunedan
5y ago
The headline is pretty misleading, as the mentioned "best practices" are highly opinionated, as the author also states: > This is my very opinionated attempt to compile some of the best practices on setting up a new Python envi
269.
▲
by
Dunedan
5y ago
In probably most cases the person who does the code changes does write tests for it as well. So as a reviewer without understanding what the code and the tests do, how do you want to assess if the tests actually do something useful?
270.
▲
by
Dunedan
5y ago
For one pet project I had to crawl a rather popular site. While that worked in general I would frequently get internal server errors. Turns out that this was the response of their CDN when it detected bot-like behavior. That left me wonderi
More ›