Re: New rx implementation with extension constructs

["Mattias Engdegård" <mattiase@acm.org>]

[Did my reply actually make it to the list? Trying again, being a fool.]

> 10 sep. 2019 kl. 19.42 skrev Mattias Engdegård <mattiase@acm.org>:
> 
> 7 sep. 2019 kl. 16.13 skrev Noam Postavsky <npostavs@gmail.com>:
>> 
>> I guess mixing function-like and variable-like definitions in the same
>> form is already somewhat Schemish, so this would just be continuing in
>> the same direction. I don't feel strongly about it either way though.
> 
> I have weak preference for the Schemeish "the thing of the left expands to the thing of the right" style, but it's nothing I'm going to war about.
> 
> Thanks for the terms 'function-like' and 'variable-like'; I will use them in the documentation.
> 
>> Yeah, since it's easy to use auto-splicing to get a seq but not vice
>> versa, auto-splicing is a clear improvement.
> 
> Right, so I switched to auto-splicing and it has no disadvantage that I can think of. Most rx constructs already have an implicit 'seq' in the body anyway.
> 
> However, there seems to be some disagreement about whether to use Lisp evaluation for macros instead of the current simple-substitution mechanism. We'll see how it turns out -- no final decision yet.