Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
algorias
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
41 ms
·
421.
▲
by
algorias
17y ago
I really liked that article, it was moving in a strange way. It was also pretty off-topic for HN, so I understand why it was killed. Submitting with the vague title "God" didn't help, either.
422.
▲
by
algorias
17y ago
Exactly. And that kind of action shouldn't be the only way of enforcing your rights.
423.
▲
by
algorias
17y ago
And that someone is getting an (arguably) fair slice of the value you produce in exchange for it. What's the problem?
424.
▲
by
algorias
17y ago
Yes, there are CUDA implementations such as flam4, but I'm always interested in new possible approaches to the problem, since I'm writing a flame editor and CUDA doesn't play all that nicely with it.
425.
▲
by
algorias
17y ago
That kind of thing also happens to me when typing. Might be related to muscle memory.
426.
▲
by
algorias
17y ago
It's possible to build good communication on top of faulty protocols. Don't blame the language. The value of the words in the dictionary stems from their widespread use. Intrinsically, they don't have any.
427.
▲
by
algorias
17y ago
It's hardly controversial for non-native speakers ;) I've always found this self-evident. Auditory learning plays a much bigger role for native speakers, and you also need to consider the proportion of reading/writing vs speaking.
428.
▲
by
algorias
17y ago
Oops, should have been more explicit. I meant fractal flames: http://en.wikipedia.org/wiki/Fractal_Flames Loved that video, though.
429.
▲
by
algorias
17y ago
Could that same logic of using shaders also apply to flames?
430.
▲
by
algorias
17y ago
> You haven't looked hard enough :-) I'm sure there's quite a bit of stuff I'd be ashamed of today. Self-awareness is key to improving. If you're not ashamed of at least a fraction of your code, you're deluded. Invariably the most talen
431.
▲
by
algorias
17y ago
It's a lot like everything. I haven't found any skill or talent where that isn't true. Keeping this in mind is the first step towards a reasonable self-assessment.
432.
▲
by
algorias
17y ago
What's the point of this inflammatory piece? Did the author speak up in the relevant forum? Others certainly did.
433.
▲
by
algorias
17y ago
What does "real" mean? When you use closures (or not-quite-closures, or however you want to call them) in python, they work just like "real" closures, except for the pathological case of defining functions in loops.
434.
▲
by
algorias
17y ago
I think it happens when the closing scope stops being local (i.e. at the time the defined function gets returned).
435.
▲
by
algorias
17y ago
Closures are the context in which a function executes. The best way to demonstrate this is with a silly example that sticks in the mind: def learn_how_to_greet(formal=True): if formal: string = "Good day, Mr. {last_name}."
436.
▲
by
algorias
17y ago
Yes, it's just a fraction of a percent. The ratio goes up if you count disposable income only, but I can't imagine it being such a big deal.
437.
▲
by
algorias
17y ago
Beaten with their own weapons: http://mug.gd/l5g1W
438.
▲
by
algorias
17y ago
But there's a difference between being able to reason about the magnitude of a number theoretically, and actually visualizing such a distance in your mind, which I find incredibly hard to do. 11 miles radius at a resolution of one pixel's w
439.
▲
by
algorias
17y ago
the "benevolent" part means he won't make any decision based solely on his own preferences, even though he has the authority to.
440.
▲
by
algorias
17y ago
The point is not whether or not he's right (which is an incredibly tricky question, actually, when you stop generalizing in a cartoonish way), it's that he's presenting something as fact which is definitely not provable.
441.
▲
by
algorias
17y ago
It's the study of worldviews (loosely speaking), not a worldview.
442.
▲
by
algorias
17y ago
"Fact 2: Nonetheless, the world is objectively real." I really disliked how the author randomly throws in opinions and beliefs like that in a context that didn't really warrant it. I thought I was going to read about epistemology; instea
443.
▲
by
algorias
17y ago
A plain text editor works for me. Copy to a word processor (or e-mail, etc) only when you get to the formatting phase.
444.
▲
by
algorias
17y ago
Hey, I take offense to that! I agree that Chile isn't exactly a tech hub, but don't generalize. Hackers and great programmers do exist here. It may just be harder to find them. Ironically, a place like hacker news helps with that ;)
445.
▲
by
algorias
17y ago
Sounds like an interesting project. I wonder, however, if writing code in this subset of python is better than directly using a compiled, statically typed language. The subset is just so tiny. No a and *k, no nested functions. Most decora
446.
▲
by
algorias
17y ago
The problem doesn't lie with the curve. It's that even those on top aren't winning big. Why play the lottery when the top prize is modest, and the risk is high?
447.
▲
by
algorias
17y ago
Scarcity doesn't add to the value of the degree. It multiplies it. The more the degree is worth by itself, the more restricting it increases its value. There's a balance point somewhere in the middle.
448.
▲
by
algorias
17y ago
trivially parallelizable processor-intensive heuristics?
449.
▲
by
algorias
17y ago
Sounds equivalent to getting rich selling the secret of success.
450.
▲
by
algorias
17y ago
5 minutes followed by a sweet aha! moment. What the article seems to miss, though (only skimmed it), is that you need to be incredibly careful with taking kwargs, specially if you're passing them along into secondary functions within the ca
More ›