After researching the topic throughly, including the way Protocols are implemented and the performance drawbacks, I don't believe my initial proposal is viable. I think the current implementation is great as-is, but I would still love to at least see warnings when comparing non literal structs.
I believe changing Access's implementation <https://stackoverflow.com/questions/33704618/why-is-elixirs-access-behavior-the-way-that-it-is> from Protocol to Behaviour had the same reasoning behind it. On Monday, October 22, 2018 at 4:46:04 PM UTC-4, OvermindDL1 wrote: > > I wouldn't recommend having the default operators delegate to such a > protocol, rather just a `cmp` or `compare` or something call, along with a > warning about using the operators on a struct if it is statically known > that it is a struct (not too uncommon). > > > On Friday, October 19, 2018 at 4:50:02 PM UTC-6, Sheharyar Naseer wrote: >> >> Struct comparisons already throw a warning >> <https://github.com/elixir-lang/elixir/blob/741dfda9f4be898c647426493f1d17fb9a4b9c53/lib/elixir/src/elixir_expand.erl#L1161-L1165> >> >> in Elixir, as it does not make a lot of sense to compare them (except in >> some very specific cases like date/time): >> >> warning: invalid comparison with struct literal ~N"2018-10-19 >>> 21:35:18.005355". Comparison operators (>, <, >=, <=) perform structural >>> and not semantic comparison. Comparing with a struct literal is unlikely to >>> give a meaningful result. Modules typically define a compare/2 function >>> that can be used for semantic comparison >>> >> >> But currently this happens only when literal structs are used. For >> example, when using `Enum.sort` to sort a list of dates, this warning is >> not raised. I know there already has been some discussion on this >> <https://groups.google.com/forum/#!topic/elixir-lang-core/eE_mMWKdVYY>, >> but I believe OvermindDL's suggested approach >> <https://github.com/elixir-lang/elixir/pull/7274#issuecomment-361362861> >> to have a Comparable protocol is much better. Kernel's comparison operators >> would delegate to this protocol, with implementations for common things >> like Date/Time, and letting it simply throw Protocol.UndefinedError for >> structs which do not have explicitly implemented it. >> > -- You received this message because you are subscribed to the Google Groups "elixir-lang-core" group. To unsubscribe from this group and stop receiving emails from it, send an email to elixir-lang-core+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/elixir-lang-core/c217dacd-0e22-4703-b0ba-29e261bba7b0%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
