I think the problems there are exacerbated a lot by over-eager type coercion and other crappy design decisions inherited from almost 30 years ago
but sometimes “👍🏽”.reverse() == “🏽👍”
the thing where it actually helps is if you’re “one word speed reading” (eg. http://onewordreader.com/). Then it’s easier to rapidly focus your eyes on each word, without having to follow a rigid timer. But if you’re reading normally it probably doesn’t help
A key part of visual design is knowing that the users don’t know what’s best for themselves. They usually stop complaining after 3 months which is proof that you are correct and they are wrong!
(sarcasm rate: 1 - ε)
many words should run into the same issue, since LLMs generally use less tokens per word than there are letters in the word. So they don’t have direct access to the letters composing the word, and have to go off indirect associations between “strawberry” and the letter “R”
duckassist seems to get most right but it claimed “ouroboros” contains 3 o’s and “phrasebook” contains one c.
are you sure there isn’t small print somewhere saying you forfeit your eternal soul to larry ellison?
is-number is a one-line function. (though it’s debatable if a function that complex should be compressed to one line)
You may have heard of a similar if more extreme “microdependency” called is-even. When you use an NPM package, you also need all the dependencies of that package, and the dependencies of those dependencies recursively. Each package has some overhead, eventually leading to this moment in time.
I tried to edit the ‘highlights’ into a single image, the top is the description of the PR, the middle is a comment replying to another comment
more accurately, average person has a higher tolerance for bullshit than for spending many hours learning something new or spending potentially years applying for citizenship in another country
I wonder if it’s common for those steganography techniques to have some mechanism for defeating the fairly simple strategy of getting 2 copies of the file from different sources, and looking at the differences between them to expose all the watermarks.
(I’d think you would need sections of watermark that are the same for any 2 or n combinations of copies of the data, which may be pretty easy to do in many cases, though the difference makes detecting the general watermarking strategy massively easier for the un-watermarkers)
C++ is std::__cxx11::list<std::__shared_ptr<table, (__gnu_cxx::_Lock_policy)0>, std::allocator<std::__shared_ptr<table, (__gnu_cxx::_Lock_policy)0> > >::erase(std::_List_const_iterator<std::__shared_ptr<table, (__gnu_cxx::_Lock_policy)0> >) /usr/include/c++/12/bits/list.tcc:158
your underflow error is someone’s underflow feature (hopefully with -fwrapv)
how did I miss that…
also a fun fact, while commercial aviation is very safe, private planes are much more dangerous, being almost as dangerous per mile as a regular car (and you get a lot more miles per hour of travel)
that works for 2 word names eg is_open or is_file, but in this case is_dialog_file_open is structured like a question, while dialog_file_is_open is structured like a statement
as many iterations as it takes
void* x = &x;
char* ptr = (char*)&x;
while (1) {
printf("%d\n", (unsigned int)*ptr);
ptr--;
}
cosmologists: sin(x) ~= 10
Considering this was written in 2001, I’m not all that worried
bonus points if you’re using a statically typed language but the library uses extensive metaprogramming seemingly for the sole purpose of hiding what types you actually need