Andreas Rumpf

> So it has to convert if into case and then back to if again? Yes.

> @Araq: please comment on "It totally sounds like the n-word, what about NIR ?" as well. En-gee-I-AR, doesn't sound like the n-word at all to me and I made...

For the sake of moving forward I renamed it to GIRN but this is ridiculous.

> Given assumeHasValue has hasValue sym, val been considered? I think it'd allow value narrowing enabling more compile time checks and evaluations. That's already covered via `nkAsgn x, value`.

> One side effect of this strategy is that finally code "blocks" look like they get duplicated in both the happy and unhappy case That's one thing I wanted to...

Maybe it works with more special kinds of labels: ```nim try: f(a) finally: echo "done" ``` becomes (roughly): ```nim TryLabel Lunused f(a) case raised of true: goto L1 FinallyLabel L1...

The backends need to handle them but it's a weird question, they exist for the backends.

I've read the paper and it doesn't really apply. It uses a new runtime and a type (or rather effect) system extension. GIRN is much simpler (to implement). GIRN is...

Somewhat related, a name like `atmdotdotatsdotdotatsdotdotatsdotdotatsdotchoosenimatstoolchainsatsnimminus1dot6dot12atslibatssystemdotnim_Init000` is a bug.

> I think this is a regression, I remember that it worked about two weeks ago, but maybe I'm wrong. I changed how cursor tracking works internally, so yes, very...