David Nadlinger

This breaks `deprecation` for Python 3.10, as `unittest2` doesn't run there (MutableMapping is now only in collections.abc).

> Can someone give a case where the return type may not be implicitly converted to immutable, given that the parameters can all be implicitly converted to immutable? That's not...

> Then I think isStronglyPure should require that the function's return type is explicitly annotated with immutable Not quite. The type must be implicitly convertible to `immutable`, as in your...

> Ping. Should I revert this to the old behaviour (allow implicit conversion to immutable) or is the new stricter behaviour preferable? Implicit conversion to `immutable` should be allowed. ```d...

Oh dear, I had a whole page-long review written up a few weeks ago, including a simplified implementation and several design comments. It seems like it never made it to...

Isn't this rather like `chain`?

More like chain(…).front with automatic application of `only` I suppose, yes. (I was referring to the more general coalesce; trying to figure out where the design sweet spot in terms...

How would a hypothetical __ctfeAlloc/Free even be related to allowing GC new in __ctfe blocks?

What is this "experience"? Immediately invoked `@trusted` lambdas are often used to emulate `@trusted` blocks, which are helpful for writing correct templated trusted code, and have also worked out nicely...

Where is the DIP for this?