On Tue, Apr 25, 2023 at 10:44 PM Jason McKesson via Std-Proposals <
std-proposals_at_[hidden]> wrote:

> On Tue, Apr 25, 2023 at 10:34 PM Barry Revzin via Std-Proposals
> <std-proposals_at_[hidden]> wrote:
> >
> >
> >
> > On Mon, Apr 24, 2023, 9:52 AM John Filleau via Std-Proposals <
> std-proposals_at_[hidden]> wrote:
> >>
> >> Floating the idea for Grouped-namespace "using" statements.
> >>
> >> A grouped-namespace using statement is a shorter way to write multiple
> using statements that share a parent namespace.
> >>
> >> using a::b::c::X, a::b::c::Y, a::b::c::Z;
> >>
> >> can be replaced with
> >>
> >> using a::b::c::{X, Y, Z};
> >>
> >> PROBLEM
> >>
> >> Sometimes I have long namespace names, and I want to alias things from
> that namespace. The current way to do that requires me to retype the
> namespace a lot. See above.
> >>
> >> This is annoying at time-of-writing because of the repetition and
> opportunity for typos. This is annoying at time-of-maintenance when
> namespace names change.
> >>
> >> SOLUTION
> >>
> >> Allow grouped-namespace using statements as seen above.
> >>
> >> BENEFIT
> >>
> >> 100% ergonomics.
> >>
> >> This solves no interesting computer science problems, nor does it
> introduce world peace.
> >>
> >> DOWNSIDES
> >>
> >> I have 0 understanding of the edge cases around this. I just know I
> want it. If other people want it, we should make it part of the standard.
> >>
> >> COST
> >>
> >> One more thing that implementations have to handle.
> >
> >
> > I like this idea. Saves on repeating the namespace, which has very
> little benefit at that point. Plus it's common for people to just avoid
> using namespace in preference to using declarations, which are quite
> verbose today due to the namespace repetition.
> >
> > Even if the namespace is short, like:
> >
> > using std::cout, std::endl;
> >
> > This kind of stuff adds up fast, and
> >
> > using std::{cout, endl};
> >
> > is just clearer.
> >
> > You can't have a { after :: today, so there's no ambiguity. This does
> make parsing a using declaration more complicated, since it's no longer
> just a sequence of qualiifed-ids, but it's probably not exceedingly
> difficult.
>
> I don't much like the use of curly braces around these things, since
> those are either introducing a scope or initializing an object. Is it
> just as syntactically valid to use `()` or `[]` in those cases?
>

Pretty sure you can't have ::( or ::[ either, so yes.

But {} has a lot of precedent in this context. This is how globs work, and
Rust does the same thing. Both support *, and both are also arbitrarily
recursive, neither of which I think we strictly need tho. One level is
probably sufficient?

Barry