TrueAsync +

[EdmondDantes]

Content

Hello!
How difficult is it to adapt the code to the TrueAsync API? What will need to be taken into account?

[binaryfire]

Hi @EdmondDantes! Big fan of the work you're doing on TrueAsync - it's cool to see native async making progress toward PHP core. Hypervel is specifically a Swoole-based framework - the Swoole runtime is deeply integrated at every level (coroutine context, connection pools, HTTP server, etc). We also use several Hyperf packages under the hood and regularly contribute back to the Hyperf framework. So we're pretty deeply embedded in the Swoole ecosystem.

Adapting to TrueAsync would essentially be a ground-up rewrite rather than an API adaptation. That's out of scope for Hypervel. If TrueAsync lands in PHP core, the more natural path would probably be for the Swoole maintainers to consider how Swoole could integrate with or provide compatibility for native async. That might be worth discussing with them directly: https://github.com/swoole/swoole-src

That said, we'd love to see the PHP async ecosystem grow - happy to chat more if you have questions about how Hypervel/Swoole handles things!

[EdmondDantes]

Yeah. I’m curious whether there’s any experience reusing Symfony packages. It would be interesting to try building something on top of a large framework.

As for asynchrony in PHP, it is seen as an unnecessary feature that complicates the PHP core. That’s the reality.
As for integration with Swoole, it’s a good idea.

[binaryfire]

I haven't used many Symfony packages with Swoole yet. Hyperf only uses their simple utility-type packages. My guess is there may be issues with state / coroutine safety in some of the bigger ones, but I can't say for sure. We use symfony/mailer and symfony/process without any issues though, so that's a good sign. Best bet is to source-dive on the packages you're interested in and look for things like static properties that might be problematic.

I plan to build a routing package for Hypervel using symfony/routing soon so I'll know more after that.

[albertcht]

Hi @EdmondDantes,

First off, thanks so much for all your hard work on the native async features for PHP core. I’ve been keeping a close eye on the True Async RFC. It’s an exciting move for the ecosystem, and I’d love to see it officially land!

As for whether Hypervel will able to support True Async down the road, I don’t have a straightforward answer just yet. Hyperf uses an "Engine" abstraction layer for its Coroutine and Server components. This is what allows it to switch between Swoole and Swow relatively easily.

Since the Swow developers are also part of the Swoole team, the way they handle coroutines is very similar, which makes swapping drivers through that Engine layer pretty seamless. If True Async’s implementation follows a path that lets us bridge it through a similar abstraction, then Hypervel could have a future there.

That said, Hypervel is currently built with a "Swoole-first" mindset and leans on some Swoole-specific features like Swoole Table. Ultimately, it'll depend on what the final specs for True Async look like and how much overlap there is with the coroutine ecosystem we’re already using with Swoole.

[EdmondDantes]

Thank you.
I once tried to build a framework for two runtimes, and it turned out not to be the best idea.
So your strategy sounds as reasonable as it gets.

As for the TrueAsync project, at the moment it will be developed as an experimental PHP fork and will receive a full API in the near future, meaning everything already described in the RFC and what was planned beyond it. How far this will go is hard to predict.