Yeah, it’s not a new idea; but I’d say “some” languages rather than “many”. Dylan’s another case; great pity it didn’t fare better. (I’d much rather be using it

Yeah, it’s not a new idea; but I’d say “some” languages rather than “many”. Dylan’s another case; great pity it didn’t fare better. (I’d much rather be using it than Swift.) Also, none of them are in the mainstream, which creates general challenges to adoption. Fortune favors the entrenched. ... Another thing worth considering IMO is structural rather than nominal typing, which fits more with the “if a value looks right, it is” philosophy that untyped language aficionados enjoy without going for full-on pants-down duck-typing. ML favors this, IIRC (I really must learn it sometime). I find nominal typing can get a bit excessively OCD, demanding so much additional annotation as to obscure the code as much as clarify (“Hi, Swift stdlib!”). I also wouldn’t mind being able to define sets of permitted type mappings in advance, and letting the compiler figure out where in the code to insert (implicit) casts. For instance, if I’m passing a variable typed as integer where a float/string is expected, it really wouldn’t kill it to promote it automatically; most are perfectly happy to promote integer literals where a float is required after all. (And if I do want to know about it, which I usually don’t, I can always turn warnings/errors on.)