Jaremy Creechley

This is still a thing. It should be easy to just specify the branch with cloning the repo.

Interesting, yah that's slower than I would expect too. Are you using release or debug? I'd also expect that ~200 elements should still be fine. It might be a good...

I had to add a 1000 items in the dropdown before it started lagging, but still it should be able to handle 10's of thousands before slowing down. Culling non-visible...

> It can be either explicit with `aref`, `{.atomic.}` or `atomic ref`, just like `Send` in Rust, or it can be inferred with an explanation when there is an issue,...

> What if you allow > > ```nim > # Package 1 > type Foo = ref object > > # Package 2 > type AFoo = atomic Foo >...

> The idea that you can retroactively claim a type with unknown refs inside to be threadsafe is pretty wild... That's a big issue with any "selective ways" to do...

> What would be really great would be "recursive smart pointers" or a smart pointer view types that enforces atomic semantics recursively: Ugghh, no that still wouldn't let you use...

> We already have --panics:on though that turn Defects into panics. The problem that I prefer to solve is that Defects are not distinguished properly: You're right that for many...

@beef331 the `RangeDefect` would be a good use cases. It's one that'd be nice to check, as converting data from outside sources could result in a RangeDefect. With `forbids:[]` currently...

> I don't know, but I assume that one can use some fixed size preallocated memory for communication with the watchdog. It depends a lot of the device. At the...