Mastodon Feed
dysfun@treehouse.systems ("gaytabase") wrote:
RE: https://mastodon.social/@deech/116986804037160449
dear types people, this is a very interesting thought and i am inclined to believe it. however, it strikes against the lore of compiler developers that i've absorbed over the years, with the conventional answer being that HKT is just inherently slow.
i'm not sure i've ever believed it was that slow, but clearly there is more to do to check the types.
my question: what can we have that will be fast and not beyond me to implement?
like i feel confident i could knock up a hindley-milner style arrangement, but i do not feel confident that i could implement a proper and useful type theory (which i would have to design first in any case...)