On 08/31/2017 10:51 AM, Zsolt Ero wrote:
> Ok, I suspected it but I'm really puzzled then why does Mike Bayer
> recommends it? Is concurrency under Python really such an impossible task?
>
> With asyncio / twisted I pretty much cannot use any of my existing code and
> any of the common libraries. They all need their own tx/aio version. No
> more import antigravity. SQLAlchemy ORM is one thing which doesn't exist
> under those worlds.
>
> Also, the second part is the programming style. If I were to write using
> that async style, I'd just write in Javascript, at least in the Node.js
> ecosystem every library is designed to work in an async environment. So you
> don't have to use ORM's with 47 Github stars
> (
https://github.com/fantix/gino) but one with 10.000
> (
https://github.com/sequelize/sequelize).
>
> So how do you solve slow HTTP endpoints with Pyramid? Just start a lot of
> workers in a server with lots of RAM?
Mike isn't a big fan of asyncio for RDBMS stuff. E.g., here is his latest
> connections. Just spent 8 hrs diagnosing a timeout due to CPU bound ops