This is a sad day for Rust. Situations can spiral out of control. Who started it? Who escalated it? You can't really define these because communication is sloppy and difficult, especially when there are cultural and language barriers involved. The only way to not get these situations to spiral out of control are a thick skin, forgiveness, and assuming noble intent.
There have been times on here where I see discussions around `unsafe` as if it was a curse. I think the discussions, over time, have gotten more nuanced which helps.
I also wonder if expectations were unclear. It now sounds like the focus of Actix was on performance and creativity / cleverness and not on being a mature product. I never heard that before. Maybe even the author didn't even originally articulate it so clearly but discovered it through these discussions.
Going forward, I hope someone archives their clones of actix and that people fork that and get it moving forward as a product. I hope people don't take a religious view of removing `unsafe` during the fork but instead follow best practices (benchmark to confirm the need, abstract it to test it, offer a feature flag for the paranoid to disable as much of it as possible).
No, it's a happy day. This won't be the last such event. People who care will learn to keep track of details they care about, and certify -- and de-certify -- without drama. Not all modules are right for use everywhere, but all are right for use somewhere, including those only trusted for personal projects.
Some of the latter will be forked, and the forks certified to higher levels, and those forked again and the forks certified or de-certified again.
This marks a step on the way to maturity, right on schedule. By the evidence, in ten years Rust will be a mature language, much more quickly than some. Most don't get there.
> I also wonder if expectations were unclear. It now sounds like the focus of Actix was on performance and creativity / cleverness and not on being a mature product. I never heard that before. Maybe even the author didn't even originally articulate it so clearly but discovered it through these discussions.
That's actually my main sticking point on this being mostly on the maintainer. From the actix documentation, releases, and promotion, I never got the feeling that it was _not_ meant for production. Even to the point of the comment "Microsoft uses this in production".
And suddenly, "It's creative/fast/research/whatever"? That really feels like trying to sidestep findings because you don't want to lose your spot in the techempower benchmarks. Even more so, when the documentation, etc. is not updated to reflect this new focus, and there's no announcements about it.
There have been times on here where I see discussions around `unsafe` as if it was a curse. I think the discussions, over time, have gotten more nuanced which helps.
I also wonder if expectations were unclear. It now sounds like the focus of Actix was on performance and creativity / cleverness and not on being a mature product. I never heard that before. Maybe even the author didn't even originally articulate it so clearly but discovered it through these discussions.
Going forward, I hope someone archives their clones of actix and that people fork that and get it moving forward as a product. I hope people don't take a religious view of removing `unsafe` during the fork but instead follow best practices (benchmark to confirm the need, abstract it to test it, offer a feature flag for the paranoid to disable as much of it as possible).