Appendix: Yet Another Introduction to Asyncio

[crazyguitar]

missing explanation.

• flow graph
• explanation

[PasqualeLivecchi]

In my opinion the following greatly simplifies explaining Asyncio.

Explanation: There are a bunch of fancy terms in asynchronous programming but they can all essentially be explained as 'context' and 'context switching'. A thread, a task, a process are all essentially the same thing, they are differently sized 'contexts environments'. Multithreading or multiprocessing is essentially context switching(sometimes between cpus). Asyncio in python is context switching for a very small 'context environment'. Tasks are slightly larger 'context environment', threads are larger than task, and process is larger than thread(and introduces cpu's). You cant have alot of threads and processes because of the context environment overhead taking up a bunch of memory, but you can have many async awaits because there is little to no overhead.

I could be wrong but its the best way to understand it quickly in my opinion.

[crazyguitar]

Thank for your suggestions. Recently, many languages introduce async/await into their syntax such as javascript. However, the hardest for me was that Python mixed generator and async/await together because Python used generator to implement a user space scheduler. That is why I created this issue for explaining how this scheduler work.

[PasqualeLivecchi]

I'm not familar with Python's use of a generator to implement a user space scheduler, but I am familiar with Pythons scheduler and have experimented with pythons async generators. Async generators are the most confusing thing in python imo. Maybe I can break it down with my context and context switching idea(ive yet to try breaking it down because my realization about asyncio essentially being context saving and context switching has only come in the last year).
yield = save context state and switch context to the calling function
await = yield from = save context state of function being awaited(yielded from) and switch context to the awaiting function
async = keyword that tells the calling function that it must be awaited(aka 'awaited' aka 'yielded from' aka context state of function being awaited is saved)
An async generator is then a function that 'async/awaits' meaning it must save the context of the function being awaited (yielded from) then switch the context to the calling function(itself) and it 'yields' meaning it must save its own context too(then switch the context to the calling function). I imagine it becomes relevant when dealing with many calls to a function that calls a function with a large loop maybe in a server or scheduler idk. I bet there are very very few mainstream libs that utilize async generators and legitimately require them.