Hacker Newsnew | comments | show | ask | jobs | submit | wmf's comments login

From that article: "Square and Starbucks are not totally splitting up. Square said it still processes debit and credit card payments at 7,000 of the coffee chain’s U.S. stores."

reply


Consensus seems like an essential property that distinguishes a blockchain from a Merkle tree, and git does not provide consensus.

reply


Yep! entirely agreed.

reply


The correct answer

reply


Imagine how much it cost to develop Servo. Now imagine rewriting every mobile app from scratch using the lessons learned from Servo. The cost would be astronomical, and for what? Incrementally better battery life?

reply


> Now imagine rewriting every mobile app from scratch using the lessons learned from Servo.

You just rewrite the popular libraries and game engines. Then lots of applications start to achieve parallel speedups. The point of Servo is in fact to achieve just that—don't forget that a browser is not just an application but a platform, running applications that happen to be written in (up-to-now) essentially-single-threaded JavaScript, HTML, and CSS. But as a specification, CSS and HTML (especially CSS) are in fact quite surprisingly parallelizable, and our tests have shown that existing apps get great speedups unmodified in Servo.

Taking the underlying framework and parallelizing it is an effective technique to take what were previously single-threaded applications and making them parallel. In fact, in a sense that's what superscalar CPUs have already been doing for decades.

reply


Most mobile apps are native GUI wrappers around ffmpeg, some http library, and maybe WebKit or Blink. The high-level coördination will remain relatively unchanged as the underlying GUI, media, network, and web libraries push up against Amdahl's law.

reply


And don't forget the king of multicore-scalable APIs: OpenGL. The reason why GPU manufacturers have been able to scale so well by adding more cores is that GLSL provides a programming model that scales broadly to (more or less) any number of cores, and applications are written to that model. This allows applications to run unmodified on new hardware with more cores and see speedups.

reply


IIRC their first experiments focused on the names of keywords and types, where they found, e.g. that "text" is a more familiar type name than "string". Presumably they'll later apply the same approach to syntax and then semantics.

reply


It seems like a reasonable approach to me; it shifts the burden of choosing the license to those who care about licensing. For example, if you work for a company with uptight lawyers you can just request Apache 2.0 and everybody's happy, rather than getting into arguments about public domain or WTFPL or some other stupid thing.

reply


But does the author actually send messages granting those licenses, or is the user supposed to "will them into existence"? Because in the first case, the author just bought himself a whole lot of stupid work, and in the second, I don't see how anyone can be sure that will hold up in court.

reply


It's a thing in the US, because "computer data" is better and more expensive than "phone data".

reply


Amazing. How do they tell whether you're tethering, though? If you're using your own smartphone, I mean.

reply


There's various ways:

Some rely on the device itself to enforce it, but that's obviously fragile if you can bring your own device.

Some check the TTL of your packets when they enter the carrier's network, because tethering is at least one more hop and so even if your computer's OS and phone's OS agree on what TTL starts at, the TTL will still be different than expected for your device's platform. Obviously this can still be mitigated by adjusting your TTL, but outside of software that'll handle this for them, that's already beyond a lot of customers.

Some even take the route of only checking HTTP traffic, and detecting tethering based on User-Agent, but I think a lot have abandoned that because it doesn't catch other protocols, and is easily bypassed even on HTTP.

reply


They don't, but on-phone software (stock iOS, as a prominent example) enforces the policy.

reply


I was furious when I bought a Nexus 7 (first version) and found out that tethering wasn't available on the device (although there was no obvious reason it shouldn't be and rooting the device would have allowed me to enable it). Running into this on stock devices seems absurd.

reply


AFAIK LDAP is a simplified version of X.400, which was part of the truly horrible ISO protocol stack.

reply


I don't think the article is suggesting NIHing the stack all the way down. As I read the article I was imagining a layer on top of SQLite that exposes a very restricted API.

reply


You're kidding but they aren't: http://0pointer.net/blog/revisiting-how-we-put-together-linu...

xdg-app is similar but with less btrfs: https://wiki.gnome.org/Projects/SandboxedApps

reply


1) I was kidding but I also thought this was possibly where some would want to take it. I personally would like to see something more Linux central and not systemd central.

2) I really like th idea of sandboxed apps. It was basically what I was saying except if there were some kind of git magic sauce in the middle of it that would be fantastic.

Unfortunately I am thinking in terms of Desktop Linux and not Corporate settings with servers. Though servers really are moving into containers.

reply


Because people think they can't use bare metal and thus its advantages are irrelevant.

reply

More

Guidelines | FAQ | Support | API | Lists | Bookmarklet | DMCA | Y Combinator | Apply | Contact

Search: