[06:42] it shouldn't [06:42] m: my num $a = 42 [06:42] rakudo-moar e78c5a7e6: OUTPUT: «===SORRY!=== Error while compiling ␤Cannot assign a literal of type Int (42) to a native variable ($a) of␤type num. You can declare the variable to be of type Real, or try to␤coerce the value with 42.Num or Num(42).␤at :1␤------> m…» [08:42] *** finanalyst left [10:51] *** guifa_ left [10:51] *** guifa joined [12:43] Then why does the first example work [12:48] m: my int $a = 42; my num $b = $a; dd $b [12:48] rakudo-moar e78c5a7e6: OUTPUT: «42e0␤» [12:48] native values auto-coerce [12:48] m: my int $a = 42; my str $b = $a; dd $b [12:48] rakudo-moar e78c5a7e6: OUTPUT: «"42"␤» [12:49] I guess the problem is that even though we have a --> int in the signature, the codegen of the call upgrades the value to Int nonetheless :-( [13:25] *** librasteve_ joined [15:35] *** librasteve_ left [15:52] ugexe: also, it looks like the optimizer makes that work: [15:52] $ raku --optimize=off -e 'my int $a = 42; my num $b = $a * $a; say $b' [15:52] This type cannot unbox to a native number: P6opaque, Int [15:53] right [15:53] the optimizer changes that to an nqp::mul_i [15:53] i actually have a branch that fixes that example. which led me to implement constant folding. which means i have to implement an optimization stage to put it in [16:18] https://github.com/rakudo/rakudo/blob/e78c5a7e672786e5ae658a50cda8515798b04ea9/src/Raku/ast/README.md?plain=1#L29 i guess we are here [16:19] although blin the ecosystem should probably be above that [16:19] yeah... [16:19] but still! [16:53] *** sjn left [18:56] *** librasteve_ joined [21:34] *** Geth left [21:55] *** RakuIRCLogger left [22:31] *** Geth joined