@owl it does recommend IEEE-754 double precision floating point numbers for good compatibility with other implementations, but things like NaN or infinity are explicitly not supported.
@owl JSON is not simple. You will know that if you have ever tried parsing or printing floating point numbers.
Also it's completely underspecified. A parser that only ever parses empty strings, empty Arrays, empty objects, null and every number as 0 is still in compliance with the RFC (although it is not following it's recommendations ...)
@owl My girlfriend uses it to mean "Lazily lying on the bed on your back with the head hanging over the edge". Essentially "lazing around". 😂
@fribbledom Apple Newton weren't smartphones since they were no phones. But I think you shouldn't discount their influences either!
@owl On one of my projects I switched to jemalloc using the jemallocator crate and tuned it to always use MADV_FREE instead of keeping the pages around.
Before this switch, glibc malloc kept 3-4 times as much memory around than was actually used and my application was constantly killed by the OOM killer.
It's absolutely crazy how much allocators can be configured and how much of a difference it can make.
Any default configuration for an allocator must be a heck of a compromise.
@owl Yeah, that makes sense. I guess the main problem is that lambdas are completely different from regular python functions and there is no inline-syntax for the latter. A questionable design decision for sure. Also generators are kind of weird in that way.
@owl Yielding from a lambda doesn't conceptually make any sense to me. Am I missing something? Maybe an example of what you are trying to do would help.
@sir Can't you bootstrap it using a binary toolchain? (if you don't trust it, cross compile the toolchain from an x86_64 system)
I propose to rename artificial intelligence to artificial Intuition,
because that's exactly what it is most of the time. A machine trained to make decisions based on intuitive "gut feeling". (Note that there are exceptions to that of course, there is machine learning that doesn't work Like that, but most of what ist called AI does)
@fribbledom Quite the opposite from my perspective. In winter the cold air condenses all the moisture from my breath and the mask gets moist quite quickly, which is very uncomfortable.
@owl Just fail and retry, then they will show something different. I usually try to trick them by clicking on pictures where I can't see what they're looking for. If you get all the others correct you will still clear the captcha. This way I'm not giving them free labor in training their AI. I now just hope that I'm not responsible for self driving cars identifying people as green traffic lights 😬.
And yes, afaik there is no way to write a trait bound for either one or the other trait being implemented, but maybe there is a way I don't know.
Everyone is welcome as long as you follow our code of conduct! Thank you. Mastodon.cloud is maintained by Sujitech, LLC.