On 8/11/2011 3:40 AM, Eric Niebler wrote:
On 8/10/2011 11:18 AM, Joel de Guzman wrote:
That example of yours [ push_back(nil(), 1) ] can also be made to work by making 'nil' random-access. In essence, we should give all sequences the highest potential traversal category that we can.
OK, let's pick a different example:
fusion::list<int> l(1); fusion::pop_back(fusion::push_back(l, 42));
Although technically this works today, it's just a curiosity due to the way joint_view iterators work.
Interesting. I wouldn't think that would work. How did that work? list is fwd, so according to your scheme
so too would be push_back(l, 42), and so we couldn't legally call pop_back on it.
Yes, but that's as far as the view concepts can proceed. Anything more will have to be a redesign of Fusion. Again, MPL's design does not fit Fusion's because of the expensive runtime operations involved with 'extensible-sequences'. So, unless you have a better suggestion, that's the best I can offer ATM. Hmmmm. now that I think about it, it is also possible to change the implementation of pop_back to make it so that it would not need to use 'prior'. The end iterator can be abstracted such that it compares equal to i through next(i). I'll see if that works out. Regards, -- Joel de Guzman http://www.boostpro.com http://boost-spirit.com