APIs and Async · Async Python · lesson 7 of 12

Running things concurrently

about 16 minutes · free · runs in your browser

await in a loop is still serial

This is the mistake worth understanding, because the code looks concurrent and is not:

results = []
for city in cities:
    results.append(await fetch(city))     # each one waits for the last

Three requests at 100ms take 300ms. asyncio.gather starts them all and waits for the set:

results = await asyncio.gather(*(fetch(c) for c in cities))

Now they overlap, and three requests take about 100ms. gather returns the results in the order you passed them in, not the order they finished — which is what makes it safe to zip back against the inputs.

If one raises, gather propagates it and abandons the rest. Pass return_exceptions=True to get the exception as a result instead and carry on.

Your turn: write fetch_all(cities) that fetches every city concurrently and returns a list of temperatures in the same order.

You start from this, and edit it in the browser:

import asyncio
import fake_httpx as httpx

BASE = "https://api.pylearn.dev"


async def fetch_temp(client, city):
    response = await client.get("/weather", params={"city": city})
    return response.json()["temp_c"]


async def fetch_all(cities):
    # Fetch every city concurrently and return the temperatures in order.
    pass