Ask HN: A New Decade. Any Predictions? (2010) • Hacker News Board

Re: Ask HN: A New Decade. Any Predictions? (2010)

#321

Post by jlokier » Fri, Jan 03, 2020, 2:51 PM UTC

bordercases wrote: post #113

zozbot234 wrote: post #62

Earlier quoted context omitted.

> Someone really needs to crack simple generic type systems with native levels of performance to get all of this to work. We know how this works, at least in broad theoretical terms. Zero-cost abstractions are quite feasible within a self-contained module/component, but their scope is ultimately limited by ease of building and deployment. Nevertheless the performance impact of having to support high levels of abstrac…

I'm thinking supercompilation might be another part of the equation as long as we're talking extended composition of modules and types whose behavior is totally closed. Having a pipeline that can supercompile when systems are pushed in production even if some compilation time is traded for performance at development (maybe "local" supercompilation for subsystems relevant to a development team?) would be an interestin…

Fwiw, I agree about supercompilation.

Some have noted that JIT systems perform many of the same things as supercompilation, and it seems reasonable to consider that JIT is part of the equation for optimising performance when systems are pushed to production. We've been doing that for decades now, with increasing sophistication in the details. Profile driven, multiple stages of specialisation and optimisation.

Not quite to the levels of proof systems (and therefore supercompilation as envisioned) yet - there's plenty of room to get better at it - and we need to go there if we want those "zero cost abstractions" across modules that aren't designed for it, often with significant impedance mismatches.

Re: Ask HN: A New Decade. Any Predictions? (2010)

#322

Post by jlokier » Fri, Jan 03, 2020, 2:55 PM UTC

School-Cotton wrote: post #98

jamiek88 wrote: post #78

Earlier quoted context omitted.

Or, like me, literally deleted my wallet with 100 ish coins in 'cos I couldn't be bothered finding a USB key. I though 'oh I'll just mine a few more next week' when I get my new laptop (!!! mining on cpu). Narrator: He forgot about it.

If you still have the old machine, it's not impossible that the deleted wallet is still on the disk and could be recovered using some sort of undelete tool.

Good thinking. I'll be happy to supply expertise recovering data from old disk images if this is the case :-)

Re: Ask HN: A New Decade. Any Predictions? (2010)

#323

Post by martincmartin » Fri, Jan 03, 2020, 2:57 PM UTC

tim333 wrote: post #195

aazaa wrote: post #185

The most striking thing about the thread is how little of current importance was even mentioned: - Bitcoin would become the decade's best investment by far - the President of the US would conduct foreign policy through Twitter - electric scooters would become a billion dollar business (Bird, Lime, etc.) - the sharing economy would threaten the taxi and hotel industries (AirBNB and Uber) - escalation of school shootin…

"C'est la vie", say the old folks It goes to show you never can tell

For those who don't recognize the reference, it's the chorus of a song made famous by Chuck Berry:

https://genius.com/Chuck-berry-you-never-can-tell-lyrics

Re: Ask HN: A New Decade. Any Predictions? (2010)

#324

Post by jlokier » Fri, Jan 03, 2020, 3:02 PM UTC

kelnos wrote: post #263

clashmeifyoucan wrote: post #256

> -People become more privacy aware after an image search engine with facial recognition is popularized and they realize that any picture ever posted of them by anyone is in the search result for their name. People become less willing to let others take compromising pictures as if they become posted, the link back to them will be made. Not an image search engine as much as whole governments but this rings eerily true…

The sad part is that I don't think this has caused people, on average, to become all that much more privacy-aware. Or, if they are, they still don't really care.

I think it's more like learned helplessness.

People don't have alternatives to using the internet, to their smartphones, and to electronic money, that let them still participate in interesting modern society.

I think deep down, anyone who thinks about it knows everything they do with their devices may be tracked, is probably being tracked and analysed, but the tools are too essential to life, so they can't afford to worry.

That said, I mentioned to someone recently that web servers keep a log of every request and the person was shocked! (Imagine if they could see the details kept by modern analytics and observability tracing!)

Re: Ask HN: A New Decade. Any Predictions? (2010)

#325

Post by swsieber » Fri, Jan 03, 2020, 3:05 PM UTC

scarface74 wrote: post #83

TulliusCicero wrote: post #54

Earlier quoted context omitted.

> - As Moore's Law marches on, dynamic languages that are even slower than Ruby are likely to catch on. They may be to Ruby what Ruby is to Java, trading even more programmer time for CPU time. Interesting how plausible this one is, yet turned out to be terribly wrong: the newer hyped languages that got some uptake were largely compiled ones like Swift, Rust, Kotlin, and Dart.

Rust and Dart hasn’t seen that much “uptake” by the broader market, it’s hyped by geeks but that’s about it. Swift is becoming popular because if you want to develop for iOS, you don’t have a choice. Kotlin is getting some uptake.

Rust has seen lots of quiet uptake. What you don't see is hiring for people for Rust because most organizations don't experience a whole-sale shift to Rust, and you just retrain people on the inside.

I'd wager there are lots of people like me - writing small tools at work in Rust, but the employer would never hire for Rust.

I think we'll see broader market adoption in the next 2-5yrs though, in terms of people actually hiring Rust developers. There's still a little maturing to do in a couple areas to clean up some papercuts.

Disclaimer: I really like Rust am a little biased.

Re: Ask HN: A New Decade. Any Predictions? (2010)

#326

Post by jlokier » Fri, Jan 03, 2020, 3:08 PM UTC

rsynnott wrote: post #318

mortenjorck wrote: post #140

> the end of 32 bit computing. By itself hardly a bold prediction, but the top reply makes a caveat about mobile devices, which gets some nods about mobile taking much longer to transition to 64-bit. Few might have imagined Apple would ship its first 64-bit phone only three years later, and a version of iOS before the decade was over that ended support for 32-bit phones!

32bit died on phones much quicker than on the desktop, in fact (though of course it died on servers first). Arguably, this should have been predictable; legacy software was a much smaller problem on phones.

I find that a little surprising, because one of the side effects of 64-bit address spaces is a lot of redundancy in pointer representations stored in RAM. 64-bit pointer-based data structures have more redundant zeros.

Much of the energy in a computer goes on communication between chips, so increased memory use means increased energy use. As well as more going on inside the CPU transferring all those redundant pointer bits around

And phones are notably sensitive to energy use.

Re: Ask HN: A New Decade. Any Predictions? (2010)

#327

Post by Joeri » Fri, Jan 03, 2020, 3:10 PM UTC

TheKarateKid wrote: post #235

geerlingguy wrote: post #125

Earlier quoted context omitted.

And PHP. PVP 5.3 was the newest version in 2010 and as the years wore on many thought HipHop by Facebook would start to take over for serious PHP applications. PHP 7 completely changed the course of the PHP community and took the air out of HHVM’s sails.

Did PHP 7 introduce optimizations similar to HipHop? I was always curious what happened to HipHop. Didn’t it become a part of Facebook’s custom flavor of PHP? (Forgot it’s name)

HipHop was a cross-compiler which facebook abandoned for HHVM, a rewrite of the PHP engine with faster internals. It always had compatibility issues, but the improved performance made it interesting (afaik nobody outside facebook really got that much into Hack the language). PHP7 was a rewrite of the internal data structures which put it on close enough performance footing. Because HHVM was never a drop-in replacement for PHP and moved further away over time the community ended up sticking with PHP7 and now pretty much only facebook uses HHVM, with the big players that moved to HHVM all transitioning back to PHP7.

Re: Ask HN: A New Decade. Any Predictions? (2010)

#328

Post by kempbellt » Fri, Jan 03, 2020, 3:12 PM UTC

Dylan16807 wrote: post #302

scandox wrote: post #275

Earlier quoted context omitted.

Well I as an innocent bystander have no idea what you mean either, unfortunately.

I'm also an innocent bystander! ryacko is perfectly clearly stating that Electron reduces the amount of training that developers need. This is different from making them develop faster. It just makes them more replaceable. I have no idea how to parse twobat's question. The "or" is especially confusing. I'm not surprised that ryacko is baffled by it.

Training is an abstract term these days. Could've meant training an algorithm, or training a person, depending on context.

Hence the training for "what" or for "who" not being immediately parseable.

Until @ryacko said "Whom do you think I'm talking about?", I wasn't sure either.

Re: Ask HN: A New Decade. Any Predictions? (2010)

#329

Post by zozbot234 » Fri, Jan 03, 2020, 3:14 PM UTC

jlokier wrote: post #326

rsynnott wrote: post #318

Earlier quoted context omitted.

32bit died on phones much quicker than on the desktop, in fact (though of course it died on servers first). Arguably, this should have been predictable; legacy software was a much smaller problem on phones.

I find that a little surprising, because one of the side effects of 64-bit address spaces is a lot of redundancy in pointer representations stored in RAM. 64-bit pointer-based data structures have more redundant zeros. Much of the energy in a computer goes on communication between chips, so increased memory use means increased energy use. As well as more going on inside the CPU transferring all those redundant pointe…

You don't have to use full 64-bit pointers for your "pointer-based data structures". You can store indexes into an arena allocation. This is common in languages such as C++ and Rust, and will become increasingly so as 32-bit is less and less used.

Re: Ask HN: A New Decade. Any Predictions? (2010)

#330

Post by scarface74 » Fri, Jan 03, 2020, 3:14 PM UTC

swsieber wrote: post #325

scarface74 wrote: post #83

Earlier quoted context omitted.

Rust and Dart hasn’t seen that much “uptake” by the broader market, it’s hyped by geeks but that’s about it. Swift is becoming popular because if you want to develop for iOS, you don’t have a choice. Kotlin is getting some uptake.

Rust has seen lots of quiet uptake. What you don't see is hiring for people for Rust because most organizations don't experience a whole-sale shift to Rust, and you just retrain people on the inside. I'd wager there are lots of people like me - writing small tools at work in Rust, but the employer would never hire for Rust. I think we'll see broader market adoption in the next 2-5yrs though, in terms of people actual…

Any citations?