Re: [boost] [MPL] Potential use for multiple sequence types

From: Emily Winch (emily_at_[hidden])
Date: 2002-08-02 15:25:47

Douglas Gregor wrote:
> On Thursday 01 August 2002 04:07 pm, Emily Winch wrote:
> 
>>So: Mike B. doesn't need a vector-based typelist (without values) in
>>order to access types in his vector-based value list. Thus, just because
>>someone needs a vector-based value list, doesn't imply the MPL should
>>contain a vector-based non-value typelist.
> 
> I see your point here, that in the example I gave there is no pressing need to 
> work on just the types in the value list. 

No, that wasn't what I meant. I meant that you don't need a type 
sequence in order to work on the types in a value sequence. The value 
sequence itself should behave like a type sequence.

This is in contradiction to my interpretation of the original statement, 
which was that the MPL needs a random-access type sequence because we 
need random access to the types of a value sequence.

>>I think I need a better set of terminology :)
> 
> I got very confused by the phrase "vector-based typelist" , but I think I get 
> it now :) Maybe "vector-based type sequence" would have helped, so that I 
> wouldn't be envisioning a cons-style typelist right after reading 
> 'vector-based'...

Yes, that was fairly confusing, sorry.

Emily.