Implement hash intrinsic

3546ce8
Opened by Patrick Walton at 2011-01-27 00:36:43

"hash" and "cmp" intrinsics need to be implemented (most likely with the native "intrinsic" mod syntax?)

  1. "cmp" actually doesn't need to be an intrinsic. The "<" operator should be sufficient.

    Patrick Walton at 2010-07-17 00:46:10

  2. We discussed this at some length earlier in the year when considering equality operators, and I think at the time we settled on having two variants. The "operator symbols" variants (<, == and such) we decided to make "shallow", only-slot-comparing, memcmp-like, not-following-boxes; whereas we were going to make "cmp" a library function (only defined on pure, i.e. acyclic, values) that does deep comparison.

    The rationale was that we wanted to make the operator cases "most like machine operations", and also that (given a deep-compare operator is much easier to specify on acyclic structures) we didn't want == to only work on immutable types, as that would be too surprising.

    I suppose we could revisit this argument.

    Graydon Hoare at 2010-07-18 15:35:24

  3. The compare intrinsic is a dupe of issue #40, so I'll make this one about hash.

    Patrick Walton at 2010-09-22 01:29:09

  4. WONTFIX (not required for bootstrapping)

    Graydon Hoare at 2011-01-27 00:36:43