ParadoxV5

MariaDB is primarily written in C++. — or are we? Hah, you will be surprised with how little C++ features we use.

MariaDB/server’s Standard Library and comparison with C++’s

my_global.h

TODO: https://jira.mariadb.org/browse/MDEV-35460

ParadoxV5

extends Timer

func await_await() -> Signal:
  print("await signal")
  await timeout
  print("await return")
  return timeout

func _enter_tree() -> void:
  print("await start")

ParadoxV5

git push defaults

Here, I will call the currently-checked-out (HEAD) local branch as simply the current branch.

The Default Repository

If the repo is omitted, git push searches the following configs and uses the first one configured:

1. branch.<current branch>.pushRemote
2. remote.pushDefault
3. branch.<current branch>.remote

The Default Refspec

ParadoxV5

Today, I realized that != and <> can be different.

Of course, they are equivalent in the ideal domain, but we live in a polymorphic world where two values can be incomparable with each other. For example, 1, a number, is incomparable to "Hello", a string of text. This relation is mathematically known as a Partial Ordering.

Arguably, "Hello" doesn’t literally equal one, which we can denote as "Hello" ≠ 1. On the other hand, "Hello" ≶ 1 doesn’t quite hold. Their mutual incomparability means that neither is less or greather than the other; it would instead be "Hello" ≸ 1.

ParadoxV5

# Basics

module Container[V, E = V]: _Container[V, E]
  include Enumerable[E]
  
  interface _Container[V, E = V]
    def self.from: (Enumerable[E] other) -> instance
    
    def include?: (V value) -> bool
    def add?: (E entry) -> E?

ParadoxV5

#include <cstdint>
#include <tuple>
#include <type_traits>
#include <iostream>
using std::size_t;

/** Compile-time Fixed-Size Array of Types */
template<typename... T> struct TypeArray {
  using TupleType = std::tuple<T...>;
  template<size_t i> using At =

ParadoxV5

#include <stddef.h>
#include <stdint.h>

/**
  Here is my Standard-compliant, SIMD-capable version of
  https://pzemtsov.github.io/2016/11/06/bug-story-alignment-on-x86.html.
  I wonder which one is faster in a benchmark?
*/
_Bool check_ip_header_sum(const char *p, size_t size)
{

ParadoxV5

#include <optional>
#include <cstdio>
int main() {
  
  std::optional<int> optional_of_none(std::nullopt);
  std::optional<std::optional<int>> optional_of_optional(optional_of_none);
  
  std::printf(
    "optional_of_none.has_value():\t%d\n"
    "optional_of_optional.has_value():\t%d\n",

ParadoxV5

puts$*[0].gsub(/(.)\1*/){'_🌷🌼🪻'[it.size]+$1}

ParadoxV5

// Lambda is a type of *value*, and *values* can’t be null.
// But y’know what can? *References* to values.

#include <functional>
#include <stdio.h>

const auto lambda = [] () -> const char* {
  return "Hello from lambda!";
};