Andreas Rumpf

As an alternative we could have ``hot``, ``cold`` and ``inline``, so today's ``inline`` would become ``hot`` and ``inline`` should always inline.

No, I cannot provide such an example (for now!) so the status quo seems sufficient. But I personally do not like non-inlined inline functions, I think I know what I'm...

That's because overloaded procs need to have the same ``untyped`` parameters in the same positions. Documented limitation. ``sequtils.toSeq`` always works then. Don't "accidentically export" a ``toSeq``.

> Wouldn't it be possible to solve the issue by avoiding an untyped implementation of toSeq in sequtils and explicitly implement it for iterators and sized collections (maybe by introducing...

> In fact even the standard library violates the "don't accidentally export a toSeq" for instance in nre, i.e. this also fails to compile I know.

Because it's in a `when` which Atlas does not evaluate: ```nim when NimMajor == 2 or (NimMajor >= 1 and NimMinor >= 9): requires "db_connector >= 0.1.0" ``` Not sure...

> And if it's not anymore, documentation on finalizers is missing. We map finalizers to destructors. Which is not entirely correct but it works well enough in practice. However, one...

This looks like a timing issue? Is it always reproducible?

That doesn't mean much beyond the fact that people avoid dependencies for various reasons.

So what does this RFC accomplish that e.g. [fusion/matching](https://nim-lang.org/blog/2021/03/10/fusion-and-pattern-matching.html) does not? You don't need to teach me about ML and Rust and Racket and Foozbaz, I know. You need to...