Madoshakalaka

Tricks: accept-word and accept-line.

In a big project, real-time AI assist usually uses the current buffer or file as a context, which is not enough information to make a perfect suggestion. But they often give good enough partially correct predictions. Once you manually provide some details, they can get the rest right.

Trick: Comment before you code.

Madoshakalaka

disclaimer: the completion setup is for vsnip but I think you can easily adapt it to other completion engines.

so the nice thing about my setup is that it only sets up html suggestion for specific projects using the power of excr files (.nvim.lua in this case). (it also works for CSS suggestions if you are into css-in-rust stuff)

First you want to have this in your lua config to enable projet-specific config files.

vim.o.exrc = true

Madoshakalaka

from typing import List


def quick_sort_find_n_th_largest(numbers: List[float], n: int) -> float:
    
    pivot = numbers[0]

    left = []
    right = []

Madoshakalaka

class ClassName:
    _instance = None

    def __init__(self, arg1, arg2, arg3):
        if ClassName._instance is None:
            # do initialization here
            ClassName._instance = self

    def __new__(cls, **kwargs):
        if cls._instance is None:

Madoshakalaka

import os
import sys
import select
import termios
import tty
import pty
from subprocess import Popen

command = 'python'
# command = 'docker run -it --rm centos /bin/bash'.split()