1. 40
    1. 25

      I’m very happy Cargo has switched to the new registry protocol. The original git-clone-everything protocol did not scale well, and Cargo managed to switch away from it before it became a major bottleneck. Crisis averted!

    2. 8

      The Cargo team removed machine-readable output from cargo test. It seems they knew this would break lots of popular IDEs/tooling, but they did it anyway? It’s concerning if they’re willing to intentionally break tooling for thousands of devs without notice and without any viable alternative.

      1. 16

        I wouldn’t characterize this as breaking.

        Before, you opted into this with -Zunstable-options, now you need RUSTC_BOOTSTRAP=1.

        In any case, this is unstable functionality, so anything goes. Unstable is unstable, if one fears breakages, one should work on stabilizing the thing.

        The more fundamental problem here is that historically libtest was an orphaned part of the project, and no one really kept an eye on tracking issues for it.

      2. 10

        I agree there should have been a transition period but I’m also glad that Rust isn’t going to be in the “we said it was unstable but people used it so now we have to stabilize it” position.

      3. 4

        Still not a good look, but epage’s comment on Reddit explains a bit:

        Once they realized it was going to break so many people, they had planned to add a transition period but that fell through the cracks until today when it was too late.

        Personally, its doing what it was advertised it’d do, be subject to breakages. The effort to stabilize it will likely see the format change.

        But yes, testing got into a “good enough state” and then not much has happened in while. I’m hoping to fix that.