Re: Re: Registries [2]

At 21:51 2003-07-30, David Abrahams wrote:
>Daniel Wallin <[email protected]> writes:
>
> > At 21:45 2003-07-29, David Abrahams wrote:
> >>Daniel Wallin <[email protected]> writes:
> >>
> >> > At 18:42 2003-07-29, David Abrahams wrote:
> >> >>Daniel Wallin <[email protected]> writes:
> >> >>
> >> >> > I'm starting to doubt our motives for local registries altogether, I
> >> >> > might be forgetting something or I might be drunk, but right now it
> >> >> > seems to me that we'd be better off with only a global registry. I
> >> >> > just can't think of any cases where there would be interest in hiding
> >> >> > converters.
> >> >> > This is quite a turnover for me, so if someone could convince me
> >> >> > that there are benefits with local registries I'd be glad. ;)
> >> >>
> >> >>Consider: one extension is converting std::vector<int> into native
> >> >>tuples, and another has exposed a class_ wrapper.
> >> >
> >> > Right, but that only justifies local registries of to_xxx converters.
> >> > Any reason to have the from_xxx converters local as well?
> >>
> >>People like to put C++ code in the global namespace and might have
> >>name collisions (?)
> >
> > Right, but with automatic importing they would be screwed anyway.
>
>Not if the automatic imports had a lower priority than the others, right?

Yes, if the local converters with higher priority doesn't match and there
is a true collision between different types. Consider:

module X
   struct A { std::string x };
   converter from lua/python string -> A
   f(const A&);

module Y
   struct A { int x };
   converter from lua/python number -> A
   f(const A&);

Y.f("foo")

---
Daniel Wallin



-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01