Predicates are incompatible with lif

[gilch]

What's the rationale for making False truthy in lif? Because that's just bizarre. I know that nil in its various incarnations is the one true false in Common Lisp. But in every Lisp I know of that actually has a false, it is, in fact, falsey.

I do understand the need for both Pythonic if and Lispy lif. There are cases it simplifies, that could fail on e.g. 0 being falsey. I think Clojure is our best guide here. In Clojure, both nil and false (and nothing else) are falsey.

Now I am not proposing a third cif. I think that over-complicates things. If there truly are good reasons for making false truthy, then we should change all the ? predicates in Hy to return nil instead of false (as with Common Lisp's p predicates). Otherwise, can we make lif act more like Clojure, please?

[agentultra]

It’s a macro and pretty harmless.

I needed a canonical lisp-ish sense of logical truth to implement a corner-case in a monadic parser-combinator I was writing to parse Hy… so that we could write READ in hy and eventually have Hy written in Hy. Python’s idea of truth is a little too broad in certain situations where 0 is a perfectly legitimate value and not a boolean.

On Jul 24, 2015, at 3:48 PM, gilch [email protected] wrote:

What's the rationale for making False truthy in lif? Because that's just bizarre. I know that nil in its various incarnations is the one true false in Common Lisp. But in every Lisp I know of that actually has a false, it is, in fact, falsey.

I do understand the need for both Pythonic if and Lispy lif. There are cases it simplifies, that could fail on e.g. 0 being falsey. I think Clojure is our best guide here. In Clojure, both nil and false (and nothing else) are falsey.

Now I am not proposing a third cif, I think that over-complicates things. If there truly are good reasons for making false truthy, then we should change all the ? predicates in Hy to return nil instead of false (as with Common Lisp's p predicates). Otherwise, can we make lif act more like Clojure, please?

—
Reply to this email directly or view it on GitHub.

[Kodiologist]

@gilch's argument seems even better now that nil is gone in place of None. That said, I never use lif myself, so I can't say I have very strong preferences here.