Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
krapht
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
20 ms
·
241.
▲
by
krapht
6y ago
Maybe Julia will become more popular, but I'd bet on the 800 pound gorilla Python getting faster through broader use of Dask, Cython, and Numba. I'll never bet against the network effects of language and library development.
242.
▲
by
krapht
6y ago
My company starts new programs in C++ because everyone knows C++ already, we write modern C++, our libraries are written in C++, and our bugs are mostly not memory safety issues.
243.
▲
by
krapht
6y ago
Of all companies poised to solve it, I think Apple could do it if they wanted. They could embed microphones in the laptop frame and integrate it with the camera as a premium feature, since they control the hardware and software stack. It&#x
244.
▲
by
krapht
6y ago
I dream of the day when our laptops come with an integrated microphone array with automatic beamforming based on head tracking.
245.
▲
by
krapht
6y ago
You speak very confidently but I would like a technical discussion of the points I raised. Even in the link you posted, there is the following: "The U.S. is not getting the message, however, because no one in the U.S. military believes
246.
▲
by
krapht
6y ago
Why are you so confident that communications jamming is easy? The person initiating the communications has all the cards; while the person trying to jam is stuck trying to blast power over a constantly shifting comms frequency and come in v
247.
▲
by
krapht
6y ago
The Great Recession was only a decade ago. Of course it's a problem, and we have recent evidence of what happens when you almost let the real estate and banking industries collapse.
248.
▲
by
krapht
6y ago
Eh, qualify fantastic. It's great in that the courses are difficult, and if you complete them it will signal you are smart computer scientist; it's backed by GaTech's name brand and the diploma is the same as the on-campus di
249.
▲
by
krapht
6y ago
This isn't really easy to solve. I'm only (a little) familiar with Chapel which I think is the most widely used high level HPC language; how data is distributed needs a lot of thought and is specified by the programmer according t
250.
▲
by
krapht
6y ago
Sure. Linear types and borrow checking is possible in Lisp too, but I wouldn't take that as evidence that it's super easy to make Clojure behave like Rust. What is possible and what is practical are two different things.
251.
▲
by
krapht
6y ago
I'm not sure that this question even makes sense. It's like asking: "What language features are needed in order to implement a borrow checker?" You don't bolt this functionality on to an existing language; it requir
252.
▲
by
krapht
6y ago
Isn't this fairly common? I've now worked at several organizations where sensitive information was stored on air-gapped networks. Software updates or data were moved in and out using pre-approved external drives. I tend to think t
253.
▲
by
krapht
6y ago
I have seen many arguments like this, as if software engineers are automata that take requirements as input and produce software code. However an important part of being in a company is developing a shared team relationship where individual
254.
▲
by
krapht
6y ago
What's whiteout for if not edit as you go? I used so much of it back in the day.
255.
▲
by
krapht
6y ago
There are limitations to MBTI; it's somewhat useful but even better would be to promote the Big 5 personality traits, which is more widely accepted in psychological type theory.
256.
▲
by
krapht
6y ago
I know lots of applications that could take advantage of 10 picosecond accuracy, if only it was feasible without spending hundreds of thousands of dollars.
257.
▲
by
krapht
6y ago
Blogs exist, they just morphed. People with things to say moved to platforms, like Facebook, Instagram, Twitter, Youtube, Twitch, or Medium. This is because there's no point in self-hosting; using a platform is free and gives you some
258.
▲
by
krapht
6y ago
I am not a game programmer, but I have read about ECS. ECS reminds me a lot more of database theory. We have unique keys (entity IDs) and values, and the goal of ECS systems is to have normalized data such that our functions that iterate ov
259.
▲
by
krapht
6y ago
I am confused. If you used to work in DSP then you know the standard practice involves MATLAB, where the default is double precision. This is also the default datatype in NumPy. Most engineers don't like working in fixed-point unless t
260.
▲
by
krapht
6y ago
Physics simulations. There's a rule of thumb that to get an n-bit accurate result after a long chain of calculations, intermediate results should be stored with 2n bits. Often using the full dynamic range of a float is necessary becaus
261.
▲
by
krapht
6y ago
Oof. Double-precision is only 2.5x better, which is less impressive than 20x for float. I still haven't found anything more cost-effective for double precision data processing (cost + dev time) than a rack full of used Xeons...
262.
▲
by
krapht
6y ago
I would ask what's going on that couldn't be solved with cygwin or virtual box.
263.
▲
by
krapht
6y ago
Reading the code of a piece of production software is a poor way to get understanding of the fundamentals of a strange topic. It's really inefficient in terms of "understanding" per hour; the student has no context as to why
264.
▲
by
krapht
6y ago
When I look at pin 5 on a schematic, it makes it easier to find on a PCB if it is physically close to pin 4 and 6. This is true in programming too. Spatially close data should be stored in memory adjacent to each other.
265.
▲
by
krapht
6y ago
http://pages.hmc.edu/harris/class/e11/lect7.pdf
266.
▲
by
krapht
6y ago
There's a huge constant factor involved, which is why we still use binary. As in many algorithms, asymptotic analysis doesn't tell the full story.
267.
▲
by
krapht
6y ago
That's not true. I was working on a system to do (not on subs, but the techniques are applicable) that a few years ago. With a big enough array, anything is possible, see the entire field of radio-astronomy. This includes sparse arrays
268.
▲
by
krapht
6y ago
Adding more states to an electronic system is trading robustness and noise insensitivity for better performance. Think about it this way: as the number of states approaches infinity, you're back to analog computing. It's a design
269.
▲
by
krapht
7y ago
As someone who occasionally has to program fpgas, the syntax of vhdl is not why performance fpga programming is difficult.
270.
▲
by
krapht
7y ago
If you wanted this today, why would you implement the Expected template instead of using Boost Outcome?
More ›