> That's a neat idea but that's not in my article. :-) But at least, I got one point right, it is written in your RFC:
_Valgrind and the Clang memory sanitizers work out of the box with Nim as it doesn 't use conservative stack tracing anymore._ However, the magic of nim still looms in: > It doesn't apply, strings and seqs have no aliases that are not statically > checked. How can this be done with a closure, eg. capturing a seq without refcounts. What is the difference between a seq and an object. I don't get it.
