HippoBaro

Limiting the amount of ram goes a long way in reducing power consumption. It takes constant voltage to keep the cells alive, even when the device is locked and seemingly doing nothing.

The iPad really does have great battery life on the other hand.

In SDR content, colors are encoded with 8-bits ranges. So, 24bits for the three color channels. With HDR, we usually use 10 bits instead (this is where the "10" comes from in "HDR10"). See https://en.wikipedia.org/wiki/Color_depth#Deep_color_(30-bit....

Yes, these are user land threads. Sometimes they are also called fibers, green threads, virtual threads, or stackfull coroutines.

I think that was an interesting read. I have worked on an userspace implementation of TCP via DPDK so I have sympathy for the limitations mentioned (the load balancing and thread scheduling argument are very accurate).

However I would have liked a section dedicated as to why hardware accelerated UDP wouldn’t be an adequate solution vs a whole new protocol. It seems to me it provides a solid basis for achieving the results the author wants to bring about?

(Context: I write async rust professionally)

Fly you fools. This will be a nightmare to debug, introspect and reason about for a speed boost that you (and your users) won’t be able to measure.

If you want to build a native app, more power to you. There are simpler languages that will enable you to do that with a much higher productivity.

Kudos to the library writer though!

You should look into Topre boards. They are very silent and have great tactile feeling. It’s not as quiet as a MacBook keyboard, but it’s as quiet as you’ll get I think

Context: I write rust professionally I have no plan of coming back to cpp.

Cpp still has the (huge) advantage of maturity. Rust is great but still half-backed. Things like thread-local-storage, custom allocators and async are either unstable or incredibly rough around the edges.

Finally, my experience has been that third-party libraries in cpp are fewer but of greater quality than the ones in rust. There are many many libraries available in rust due to the amazing dependency system. But the average quality is quite poor.

In cpp you tend to be conservative when adding dependencies which means you have a much better understanding of the code you run. Or, you write your own code that’s usually custom made for your purpose, which generally end up as being more performant.

POSIX is not linked to C. POSIX is a specification and that can be implemented in any language.

What POSIX is is outside of the control of Linux. Linux wants to be POSIX compliant for compatibility. Changing POSIX would not make sense.

This is great, and we should celebrate China for doing this. I would genuinely love to see this kind of legislative will in Western society.

My main question is that it seems very hard to distinguish between a legitimate recommendation system and a system that is designed to capture your attention. For instance, I like being able to see related items on Amazon when evaluating a purchase. That recommendation empowers me as a consumer. China can afford to make the law vague because the regulator will never be challenged in its interpretation. That wouldn't work elsewhere, I think.

That being said, not all is great. China grants rights to people over their data while making it clear that they don't own any of it. This legislation is part of a broader regulatory framework designed to give the state a true monopoly free of all limits. This should give us pause.

So UUIDs were the perfect solution except that, if a user double-click a url with a UUID in it, it won’t select all of it.

Can’t you just… remove the hyphens? UUIDs are just arrays of bytes, the hyphens are not part of the data.

Instead, they use a less popular, less supported, more collision-prone ID format.

I mention io_uring in my original comment. The issue isn’t that the tech doesn’t exist, it’s that it’s not being used.

And even before that there was Linux AIO etc.

You can implement stuff using eBPF and do IO and compute in without leaving the kernel. It’s crazy.

For instance you can perform an on-disk b-tree lookup in the kernel.

These devices are so fast that crossing the kernel boundary becomes the limiting factor.

The problem with fast byte-addressable non-volatile memory isn't hardware IMO. They've been around for a while, with Optane DC being the better option at the moment.

The real issue is that software and APIs aren't keeping up with the hardware. If you do IO with memory mappings or anything POSIX, you will not be able to saturate a modern NVME SSD. And I'm not even talking about Optane here.

This means that there is no market for these crazy fast devices. Look at the storage-optimized instances of every major cloud provider, and you won't find them. They would be too expensive, and the software people would run on them wouldn't be any faster than the older, cheaper SSDs.

It'll take time for the software to evolve and catch up. Initiatives like DAX, eBPF, and io_uring in the kernel are a big step forward in the right direction.

My company wants to bring me over to the US from France under an L1-B visa. The pandemic makes getting interviewed at an embassy very difficult, and the process has stalled because of this. Do you have any idea when visa processing will resume, and is there anything my employer or I can do to speed things up?

I’ve been using this for years and it’s been extremely useful. The feature to plot the prices over time is great to gauge whether it’s a good time to buy something or if you’d rather wait.

That explains the Chinese economic miracle quite well!

> Historically software was always low on memory, and subsequently sorting algorithms were developed that bent over backwards in order to use as little additional memory as possible. Nowadays most modern systems have several gigabytes of memory that are not used and are just sitting idle.

True, but on these systems, most of the time, users don't care about performance. And when they do, the dataset is not large enough for this kind of algorithms to shine anyway.

I spent two years in Tsinghua University doing my master's degree a couple of years ago. Even though Chinese universities never enjoyed the freedom of speech/research/though as in the west, putting it on paper is very sad.

From what I understand, they largely romanticize the times when Hong Kong was a British colony. However, it seems clear that what the protesters are asking for is ideologically more in-line with a British/Western conception of a democratic society. And because Hong Kong will never be independent and the CCP never democratic, raising the British HK flag is the last thing they can hold on to.

I see it as a testament to their desperation.

Go programs are statically built as fat binaries. That basically solves the problem in itself. Python scripts rely on a (very) large shared constellation of thing being present on the host system. Change one of these and you might break some stuff.

The reason is that HTTP/2 will be short-lived. Most of its potential (protocol multiplexing, mainly) is wasted because it still runs over TCP/IP. HTTP/3 will correct that with QUIC. I think we can except HTTP/3 to really see widespread adoption. HTTP/1.1 will remain ubiquitous though because there's a gazillion box that only speaks that.

This. When you think about it, it's an exciting time to be a dev. We need to be clever at stuff and can't just expect next-generation CPUs to make coffee for us.

Sure, you can. But then you need to manage when GC triggers. Because if you don't, you'll drop frames. And then the real choice is between C++ where there is (little) extra work getting RAII correct vs GC languages where you need to keep preventing GC up to a point where it's okay to freeze the universe.

I'm working with C++17 right now and I personally can relate to the author. Reading the discussions around C++20 has been an eye-opening experience for me. C++ was already a difficult language but has become a behemoth of complexity. But it has to be said that once you know "Modern" C++ it's a very productive language.

Somehow, it feels good to use it and after a while you find yourself reveling in complexity. I'm almost ashamed to say really.

You're onto something!

Probably more because Linus is pretty much bs-free (for the better and worse).

Thanks for the info, couple of things I didn't know. What I meant by "it's not a VPN" was mainly it doesn't make your computer part of another private network.

Shadowshocks is not a VPN at all. It's a connectionless SOCKS5 proxy. No hand-check, no key exchange, no protocol agreement. VPN is not made to be undetected, only secure in the sense of data encryption. Shodowsocks has broader objectives.

See https://shadowsocks.org/ for info.

(I live in China) I would advise against VPN entirely. I am unsure about the state ability to decrypt the content of the connection (heavily depends on how the VPN is configured really — weak and legacy ciphers, etc.). But they will detect it and eventually you'll start dropping packets like crazy.

A simple way to evade all of this is to use shadowsocks with a strong cipher and strong password between your computer in China and your server outside of it. Don't use any free server and don't use any commercial shadowsocks offerings. Set it up yourself, it's pretty easy.

On the mobile phone side of things. I wouldn't trust anything. Especially Apple that has been very complaisant with local authorities.

China plays a tactical game: they pretend (or we suspect, and they want us to) that they can do a lot of things. But nobody knows the extent of what they are actually capable of.