Today they still do it in python except it’s now running on Spark across sprawling expensive clusters to do complex tasks like transform text or coalesce two fields.
> where you have to use countless trivial unit tests to have any confidence about what your code is doing.
I don’t understand this. I can read well written code and tell you what it does, including in old school scripting languages. I don’t need countless trivial unit tests.
But that negatively impacts the revenue of the various cloud providers and puts devops engineers back into the same unemployment line when their job title was sysadmin.
Figma is written in C++ and compiled to wasm, and it's amazing, you're right.
But that approach represents maybe 0.001% of modern front end development. The curent popular path is React/Vue and the endless JS toolchains to compile these horrible bloated broken apps.
fyi on swoole, this book [1] just came out (I cannot comment on the quality, haven't read yet) which hopefully will help bridge the English documentation gap. I've been using swoole for a couple of years and google-translating my way through the Chinese wiki and git issues/comments/etc.
Swoole (module) is extremely performant in anything networking, including as a websocket server. There are other php-only options (amphp/websocket-server, ratchetphp/Ratchet) but if performance is your need, stick with swoole.
I'm not talking about the quantity, I'm talking about a list of who upvoted and downvoted a specific comment/post.
Also what you've described happens anyway. The fact that a number isn't displayed doesn't really matter when comments are ultimately ordered by score. When someone writes a comment that doesn't adhere to the current groupthink, they get downvoted. When people are downvoted too much they delete their comments, stop commenting, or through pavlovian reinforcement change their opinions until they get positive points.