On Fri, Dec 14, 2018 at 9:51 PM Michael Jones <michael.jo...@gmail.com>
wrote:

> There is no absolute reason why a, b, and c could not be evaluated at
> compile time.
>

Being pure is not enough for that. You need at least pure and total -
otherwise you could have

pure func Foo(x int) int {
    for {}
    return x
}

However, (dis)proving totality is not possible (not even touching on the
fact that totality alone isn't even enough, when you also want speedy
compilation). So to actually get this to work, you'd need to restrict pure
functions to a non-turing-complete subset of Go (that is still useful),
which is pretty complex to do.

Especially given that the cases you mention (and all the cases I can think
of where I'd definitely want that to happen) are just inlining + constant
folding (+ maybe common subexpression elimination), Ians suggestion of
keeping this a compiler optimization seems more sensible.


> There are many practical reasons why it cannot be done today.
>
> If a function was labeled as pure ("pure func ...") the compiler would not
> even need think hard, and if purity were a reflectable attribute, then it
> is imaginable that compiling a function invocation could be:
>
> "if package is a standard one (like math) and the function is a pure one,
> then call func on argument and use the result here"
>
> Doing this for user packages brings up deeper issues and is harder.
>
> --
>
> *Michael T. jonesmichael.jo...@gmail.com <michael.jo...@gmail.com>*
>
> --
> You received this message because you are subscribed to the Google Groups
> "golang-nuts" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to golang-nuts+unsubscr...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>

-- 
You received this message because you are subscribed to the Google Groups 
"golang-nuts" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to golang-nuts+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to