Re: [boost] [Containers] Review

Subject: Re: [boost] [Containers] Review
From: Phil Endecott (spam_from_boost_dev_at_[hidden])
Date: 2011-08-11 10:10:08

Nathan Ridge wrote:
>> Regarding the batched insert, we have previously discussed the
>> complexity of inserting a range. Fundamentally, however, I don't think
>> it is acceptable to have efficient insertion ONLY in the case of the
>> insert(begin,end) method: it is too common to want to insert several
>> values in a loop without creating a temporary container. I'm not aware
>> of any established (i.e. std:: or boost::) pattern for doing this:
>
> There is no need to create a temporary container to use a method that
> takes a pair of iterators. With the help of the Boost.Range library
> and some lambdas (Boost.Phoenix or C++0x, depending on your taste
> and constraints), you should be able to create a lazy range view
> of wherever you're getting your data from in your loop.

Well, yes and no.  If I show this code to a Java or C programmer, they 
can probably work out what it does very easily:

int i = 0;
while (i != 10) {
   container.insert(i);
   i = f(i);
}

How easily would they understand your version?

There are also cases where it is fundamentally hard to write a range 
generator without using e.g. co-routines:

void towers_of_hanoi(container& c, string from, string to, string tmp, 
int n)
{
   if (n==1) {
     c.append(make_pair(from,to));  // "Move a disk from " from " to " to."
   } else {
     towers_of_hanoi(c,from,tmp,to,n-1);
     towers_of_hanoi(c,from,to,tmp,1);
     towers_of_hanoi(c,tmp,to,from,n-1);
   }
}

I really don't think it's acceptable to require users to use tricky and 
limited features like lazy range generators just to get efficient 
insertion into a container.

Regards,  Phil.