Re: [boost] [extension] factory_map::get() suggestion

From: Robert Dailey (rcdailey_at_[hidden])
Date: 2008-07-03 20:49:42

On Thu, Jul 3, 2008 at 7:36 PM, Jeremy Pack <rostovpack_at_[hidden]> wrote:
> Steven,
>
> Yeah, it isn't quite that simple. I read the original post too quickly and
> thought he was referring to a different function.
>
> I originally had a type for that std::map<int, factory<word> >, so it would
> be something like type_factory_map<int, word>, which isn't much terser.
> (thus, I removed it)
>
> I think the best solution here is to make your own typedefs for your
> specific case.
>
> There are a couple of other places in the library though where I intend to
> add the type of typedefs suggested.
>
> Jeremy
>
> On Thu, Jul 3, 2008 at 5:26 PM, Steven Watanabe <watanabesj_at_[hidden]>
> wrote:
>
>> AMDG
>>
>> Jeremy Pack wrote:
>>
>>> Yep! Good idea. I'll take care of it.
>>>
>>>
>>>> For example, right now I have to do this:
>>>>
>>>> factory_map fm;
>>>> std::map<int, factory<word> > & factory_list = fm.get<word, int>();
>>>>
>>>> It would be ideal to do this instead:
>>>>
>>>> factory_map fm;
>>>> factory_map::type& factory_list = fm.get<word, int>();
>>>>
>>>>
>>>
>> Is it really that easy?  Doesn't the result of get
>> depend on the particular template arguments?
>> I don't quite see how you can get away with a
>> single typedef.
>>
>> In Christ,
>> Steven Watanabe
>>
>>
>> _______________________________________________
>> Unsubscribe & other changes:
>> http://lists.boost.org/mailman/listinfo.cgi/boost
>>
> _______________________________________________
> Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
>

Why not turn factory_map::get() into a template utility class of some
sort instead of a function? That would give you the ability to typedef
the std::map. Perhaps another option is to do the following:

factory_map<word, int>::get()

I don't have time right now to give any reasonable ideas, but those
are just some thoughts off the top of my head. Later on I may try to
help find an alternative to this. All I know for sure is that exposing
std::map seems like an implementation detail, and specifying the same
template parameters more than once also bothers me slightly.