jepler

43 ms·

421.
by jepler 9y ago
thanks for that
422.
by jepler 9y ago
wow that totally fails to explain what nameof is. It seems to be a tiny bit similar to the C preprocessor's stringize, #include <boost/preprocessor/stringize.hpp> #include <stdio.h> int main() {
423.
by jepler 9y ago
Sounds too good to be true. Who are these people and why should we trust them with our data? Page title is in cyrillic script. fex.net's IP address is in a Ukraine block, with one contact e-mail in .ua and one in .bg. Contact addres
424.
by jepler 9y ago
Sure, it was for decades. FPATAN in the x87 instruction set computes the 4-quadrant arctangent.
425.
by jepler 9y ago
brutal takedown http://www.funraniumlabs.com/2017/08/laser-products-hate/
426.
by jepler 9y ago
Sure: CPython doesn't know statically that anything is "not an object", so your proposal adds an extra test to every incref and every decref. This is perceived as probably worse for performance. In 2002, a hacked CPython in
427.
by jepler 9y ago
and having allowed myself to read the rest of the replies .. fizz bzzzzt!
428.
by jepler 9y ago
Hm. I don't quite see how to do it, but I think I know one of the tricks you have in mind. If you have the book Hacker's Delight handy, open it up to 10-17, "Test for Zero Remainder after Division by a Constant". With
429.
by jepler 9y ago
maybe not twice as fast, but my favorite open source program's ./configure runs in 2.43s with the taskset hack (best of 8) and 3.40s without it (best of 8). That's just a 28% improvement, but it's enough to make me cons
430.
by jepler 9y ago
Probably a new entry for https://en.wikipedia.org/wiki/Riemann_hypothesis#Attempted_p... but one can always hope an attempted proof will withstand scrutiny.
431.
by jepler 9y ago
well suited for decimal money, though to be fair that's just about all money these days. Adding or subtracting numbers with any fractional part hits the "slow path". A slightly more clever implementation could add "eq
432.
by jepler 9y ago
Yes, but at the same time you'd expect each library to be used by many applications. Say you have 100 libraries, you can make 10,000 applications just by picking 1 or 2 of those 100 libraries, applications should far outnumber librari
433.
by jepler 9y ago
I only wish that at $DAY_JOB I ever encountered a performance problem that was as subtle as "oh you're not properly using the L2 cache". It's more like, "why did you choose an N^3 algorithm when you can do it in N
434.
by jepler 9y ago
To actually be serious instead of flip, I think "Calvin" is just the name of the software, without being an acronym or initialism. However, the authors may be alluding to John Calvin and perhaps some Calvinist idea such as predes
435.
by jepler 9y ago
CALVIN: Calvin acronym lacking verifiable industry nomenclature
436.
by jepler 9y ago
wtf mozilla. (verified here, open addons and click "get add-ons")
437.
by jepler 9y ago
yes.
438.
by jepler 9y ago
.. only tested with gnu tar, but all the world's a vax eh?
439.
by jepler 9y ago
I went and researched this before reading the link, and found a slightly different solution than the author. By taking advantage of python's 'tarfile' module, it was easy to create a (us)tar-compatible header. My solution i
440.
by jepler 9y ago
.. some haskell implementations apparently had (practically) limited stack size, at least as of 2007, so as long as "pure" functions still allocate stack frames, it's possible for the result of the pure function call to be &q
441.
by jepler 9y ago
Using an undefined value leads to undefined behavior under the C spec. This "undefined behavior" could be a visible side effect (in the helpful case, a diagnostic on stderr like when you're using valgrind on your program; in
442.
by jepler 9y ago
I doubt pure functions exist in any language. For instance, consider the C function 'unsigned twice(unsigned a) { return a + a; }'. Here are some ways it can be non-pure: It could be called with an argument that has never been g
443.
by jepler 9y ago
this is something to do with the "triangles/cone" tunable. Changing from the default 50 to 48, the vertical line becomes actually vertical.
444.
by jepler 9y ago
Let's start with caloric requirements. I found http://www.waldeneffect.org/blog/Calories_per_acre_for_vario... which I'll take at face value. An acre irrigated and planted with potatoes supplies 17.8 millio
445.
by jepler 9y ago
Python has it: 'del' C++11 has it, as long as your intermediate thing is not a parameter: auto real_var = []() { complicated lambda; return calculated_value; }(); I'm sure someone will figure out how to polyfill it into javas
446.
by jepler 9y ago
Another flaw with generate_canonical<float> is that if you generate, say, 2^40 of the things, you should expect to get quite a few values smaller than 2^-32; but with minstd_rand, or mt19937 I believe you will not ever see values smal
447.
by jepler 9y ago
C++11's random number facility isn't as good as all that. Here are some of the pain points I'm aware of: There's no way to use the random device to uniformly seed the internal state of an arbitrary pseudorandom generator
448.
by jepler 9y ago
I am not a specialist in GSM networking, but I went looking at the relevant standards documents to see whether I could answer your question. It appears that the handset can request information from the SIM about frequency bands. For exampl
449.
by jepler 9y ago
A good high-level overview for those new to the topic. Here's my advice: If you have the time to do error analysis and performance optimization, then maybe use float; if not, use double. http://www.lsi.upc.edu/~robert
450.
by jepler 9y ago
As someone who has been considering a Ryzen-based box for the assumed lightning speeds of parallel compilation, I'll be following this story.

More ›