Re: On elisp running native

[Andrea Corallo]

Spenser Truex <address@hidden> writes:

> Andrea Corallo <address@hidden> writes:
>
>> In case this is of interest I'd love to see it as feature branch to
>> better collect feedback and work on its evolution.
>
> I have a comment about `1.6 Compiler hints' 
> http://akrl.sdf.org/gccemacs.html#org7d51b3b
>
> I think it is better to avoid using separate symbol for each hint, and instead
> use on that looks like a normal prefix notation.
>
> (comp-hint-cons x)  --> (comp-hint cons x)
>
> This keeps the namespace cleaner. Also, consider the situation of specifying
> multiple types, which may be desired at some point:
>
> (comp-hint (cons fixnum...) x)
>
> which can't be done symbolically
>
> (comp-hint-cons-fixnum-...? x)
> (comp-hint-fixnum-cons...? x)
>
> All this compiler stuff is way over my head, so if my syntactic taste is
> overshadowed by a more important consideration I will not be surprised.

Hi Spenser,

two thoughts on this:

- `comp-hint-cons' and `comp-hint-fixnum' are imagined to be produced by
  macro expansion and not meant to be used directly by the user.

- As of today the compiler is not that smart to take advantage of
  something that can potentially more then one type and I don't know if
  it will ever be, at least for the code generation side.  Perhaps could
  be useful for code analysis and diagnostic.

That said you are right this is not really flexible or general and I
think your suggestion is correct.

Thanks!

  Andrea

--
address@hidden