Ian Douglas Scott

Good question. I was thinking of types like `String` that involve allocation and can't be constructed in a const context. But that doesn't seem to be a problem. `const fn...

Oh, bigger issue: `Default::default` is (unsurprisingly) not a `const fn`. So presumably a lot of uses of this crate break it it were always `const`. That happened to not be...

Using `#[new(const)]` from this branch when there's a `#[new(default)]` field fails to compile with: ``` error[E0723]: can only call other `const fn` within a `const fn`, but `::default` is not...

Looking at https://github.com/nrc/derive-new#examples, `#[new(value = "vec![1]")]` would presumably also fail to compile as a `const fn`, and while `#[new(value = "42")]` would be fine. Which is not really possible for...

I think that should work. It is potentially a backwards compatability issue though. (The Readme specifically mentions adding a new member without breaking comparability as a goal of the crate.)...

I've also noticed that the API lacks something like this. I guess changing it into an owned value doesn't really have the expected semantics of a `Clone`, `to_owned` means something...

Or... I see you already mentioned that same crate on https://github.com/jnqnfe/pulse-binding-rust/issues/44. There is a bit of complexity in `SinkInfo` since it has `Option` fields and a `Vec

https://en.opensuse.org/HCL:Pinebook-Pro-RK3399 suggests some sort of issue with 2021.10 on the PBP: > Screen is not working with upstream u-boot version 2021.10. This is fixed in u-boot 2022.01

> If we decide we want end to end tests of each DNS provider, I don't think we want a generic plugin based on Lexicon because we cannot reliably test...

Looking at `pygobject`, this seems to be done with `gtk_widget_class_set_connect_func` on GTK3 and `GtkBuilderScope` with GTK4. > Could maybe make use of https://crates.io/crates/label for this That uses the `ctor` crate....