Re: [boost] lifetime of ranges vs. iterators

AMDG Arno Schödl wrote:

So any implementation that even temporarily expands iterators to their original size, like Giovanni's implementation, will need vastly more storage. IMO, relying on the compiler to fix this is hazardous. Non-optimized code should be a constant factor slower than optimized code, and need a constant factor more memory, not exponentially slower and exponentially more memory.

If I understand Giovanni's implementation correctly, it only needs to expand one layer at a time. In Christ, Steven Watanabe