Boost logo

Boost :

From: Yitzhak Sapir (yitzhaks_at_[hidden])
Date: 2002-08-08 10:06:25


There are only 3 or 4 algorithms that receive an _n argument though many times I've wanted to do a copy using an n or bound some other algorithm for n elements. In the case of strings the problem is solved since I can look up the function in string's many function interface, but many other classes and range types do not provide me this freedom. So, my question is: Is there such a thing as a bounded_iterator (for lack of a different name)?

My idea of a bounded_iterator would have an interface similar to filter_iterator, except it gets a difference_type bound. It maintains count through ++, --, and += of how far you are from the bounds, and any point beyond the bounds is considered end.

This way doing:
        std::copy (make_bounded_iterator(str.begin(), str.end(), 5), make_bounded_iterator(str.end()), std::back_inserter(copy_of_str));
would be equivalent to
        strncpy (new, str, 5);

Is there such a thing? (If not, is there any interest in having such a thing?) I tried to look on the site but didn't seem to find something like this, although using counting_iterator in some algorithms might give a similar effect. Perhaps using counting iterator with a filter and std::less might do something like it?

Incidentally, while I'm having problems using filter iterator in VC6, I think perhaps interface wise, it would be able to do make_filter_iterator(end_iter) to construct an end iterator.

My idea of semantics are as follows:
iter = make_bounded_iterator(begin, begin + 7, 5)
would act as follows, if the following are executed in sequence:
++iter --> iter now points begin + 1
iter += 5 -> iter now seems to point at begin + 7, because iter was begin + 1 to begin with, and adding five would have moved iter to begin + 6. Since that would have been out of bounds, the internal iterator is made to point at begin + 5, but the initialized "end" pointer is visible in this case -- so iter returns begin + 7. (Alternatively, this can be defined to be an invalid operation as is the following:)
++iter; --> this is undefined. iter points at end, so you can't increment iter.
--iter -> assuming this was executed after iter += 5, iter now points at begin + 4, since the internal iterator was pointing at begin + 5 to begin with. The internal iterator was moved back to begin + 4, was found to be within the range again. Therefore the the internal iterator is now visible as the value of iter.


Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk