Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
mratsim
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
28 ms
·
301.
▲
by
mratsim
6y ago
Looks very interesting but ... AVX512 is already problematic cooling wise, this seems even worse.
302.
▲
by
mratsim
6y ago
"The first truly global event", I expect the World Wars or the Olympic games to challenge that in the past century. And the conquest of the sea and riches beyond that.
303.
▲
by
mratsim
6y ago
3.5% of the population needed apparently (Academic study): - https://www.bbc.com/future/article/20190513-it-only-takes-35...
304.
▲
by
mratsim
6y ago
At this point, if a company raised an SIO (Softbank Initial Offering) and starts with a W like WeWork and Wirecard, I'm going to look into it at least a hundred times or avoid it like the plague.
305.
▲
by
mratsim
6y ago
Companies have been using Lisp to accelerate game development with a scripting language in AA game. Making a game is a hard problem. See Lisp on Playstation 2: https://en.wikipedia.org/wiki/Game_Oriented_Assembly_Lisp
306.
▲
by
mratsim
6y ago
Intel vs AMD in the past is a typical example on how money + exclusive deal could almost kill off a competitor. Then Intel would enjoy a monopoly and instead of being stuck at Skylake for 4 years, we may have been stuck at Core 2 for 10 ye
307.
▲
by
mratsim
6y ago
When current teachers retire or the market (unemployed students) forces them to adapt the curriculum.
308.
▲
by
mratsim
6y ago
That's called the Expression Problem, there is no free lunch.
309.
▲
by
mratsim
6y ago
Ah yes, I would love to recompile and run that through libFuzzer and collect all the hidden bugs that were never discovered when it was a blackbox.
310.
▲
by
mratsim
6y ago
What? So what do German use for 10^9, Milliarden? In French we use Milliard for billion and Trilliard for Trillions (and millions for millions).
311.
▲
by
mratsim
6y ago
I prefer in-depth article that don't skip over "details" and I expect that from an university publication. Otherwise I would read my Twitter feed.
312.
▲
by
mratsim
6y ago
> While this is ultimately a political problem it will only be practiced where it’s enforceable like all other political problems. Building a company in the US becomes an untenable legal and security risk as if you don't backdoor y
313.
▲
by
mratsim
6y ago
But clever technological workarounds might expose those idiots for what they are: Clowns trying to "regulate" things they don't understand without seeking expert advices beyond their narrow lobbyists.
314.
▲
by
mratsim
6y ago
The 4 horsemen of the infocalypse: - terrorists, - drug dealers, - pedophiles, - organized crime. Quoting https://en.wikipedia.org/wiki/Four_Horsemen_of_the_Infocalyp... Have a target "thing" you wish to stop
315.
▲
by
mratsim
6y ago
Given how controversial gun ownership is, I wouldn't use that as an example, it might undermine the case.
316.
▲
by
mratsim
6y ago
I guess it's time for github.eu
317.
▲
by
mratsim
6y ago
What if you put a text in your project: All lawsuits are in the jurisdiction of X Tribunal, situated in Switzerland or another privacy-friendly country?
318.
▲
by
mratsim
6y ago
This would be in breach of various privacy laws in the EU. Companies have been sued for using weak encryption (cough plaintext passwords).
319.
▲
by
mratsim
6y ago
Banks will love that. As would EU, South American and Asian businesses, they would VPN to another continent VPN to finally have nested encryption for key stakeholders communication.
320.
▲
by
mratsim
6y ago
Then if I'm writing a concurrent queue or concurrent hash-table I'm on my own since 100% of my codebase is unsafe Rust. Which was exactly my point. Rust provides no support for fearless concurrency because actual concurrent code,
321.
▲
by
mratsim
6y ago
Ah yes, I watched that talk at FOSDEM 2019, it was very nice.
322.
▲
by
mratsim
6y ago
Yes low-level C + retaining memory around in the GC. The second part avoids many malloc/free that may be hidden in destructors in low-level languages.
323.
▲
by
mratsim
6y ago
> In these cases we want the properties of OOP: local retention, message passing, abstraction through public interfaces and state-machine-like behavior. That's not OOP but Actors or event loops.
324.
▲
by
mratsim
6y ago
You can write procedural without having to resort to manual memory management. C++ and Nim (for citing languages I know the best) gives you tools, destructors for both and for Nim a tunable/deactivable GC, that you allow you to write p
325.
▲
by
mratsim
6y ago
My rule for writing OO code is to not write OO code. Procedural code is: - simpler to analyze, audit and (formally verify) - simpler to multithread - maps to what a program does: a series of transformations applied to an input - maps to low
326.
▲
by
mratsim
6y ago
Banks are very creative in their OTC derivatives.
327.
▲
by
mratsim
6y ago
They surely did, and got the very unique Software Transactional Memory based multithreading.
328.
▲
by
mratsim
6y ago
> For example, Hindley-Milner languages cannot add sub-typing Subtyping? Haskell has ADTs.
329.
▲
by
mratsim
6y ago
Nim JSON parser is not optimized, it was mostly written for maintenance (for example it allocates a Table per node in the json file) I.e. the difference in speed here is a difference in elbow grease. I have yet to come into an optimization
330.
▲
by
mratsim
6y ago
So Crystal is trying to beat C++?
More ›