On Mon, Mar 21, 2011 at 12:42 AM, Vicente Botet <vicente.botet@wanadoo.fr>wrote:
Max Sobolev wrote:
On 19.03.2011 23:12, Vicente Botet wrote: > > Jeffrey Lee Hellrung, Jr.-2 wrote: >> >> Then all I can suggest as a better name is *maybe* scoped_type
>> scoped_enum_type. But scoping_type might be fine. >> > > Hmm, scoping_type and scoped_enum_type seems reasonable, but I
or prefer
yet > the use of native_type. What do others think?
I like scoped_enum_type name :)
Sorry, I think I misunderstood Jeffrey.
were you proposing to replace scoping_type by scoped_enum_type or or scoped_type and let native_type as now?
The quote referred to replacing scoping[_type] with either scoped[_type] or scoped_enum[_type] (I like the latter better). I had already commented that I think native_enum[_type] would be clearer than native[_type] and underlying_integer[_type] clearer than underlying[_type] (although this latter case is already called underlying_type in the standard, right?). Robert Stewart brought up a good point on the redundancy of the "_type" suffix, so although the above alternatives to your current metafunction names are longer, that will be mostly mitigated by removing the redundant "_type" suffix. However, we do have free function analogues of these metafunctions, correct? Maybe you can put the metafunctions in a result_of namespace, as Boost.Fusion does? [I'm hoping the angle brackets get preserved...] namespace result_of{ template< class E > struct native_enum; template< class E > struct underlying_integer; template< class E > struct scoped_enum; } // namespace result_of template< class E > typename result_of::native_enum<E>::type native_enum(const E e); // etc. - Jeff