Paolo Carlini <[EMAIL PROTECTED]> writes: | Hi Gaby, | | >There is something that we knoq about the code, that the compiler | >does not have: we expect a sequence denoted by its first and last | >element, so that the difference is always positive. However, the | >difference of pointers is defined in the language as an ptrdiff_t (a | >signed type). No wonder what people invented ssize_t :-) | > | >I believe we have to investigate further what promises we want make | >and hold through max_size() and change the code according to that. | > | First, I want to tell you that I agree with your general point of | view, here and elsewhere, in particular about the tension existing | between difference_type and size_type: as a matter of fact, when, time | ago, I noticed some differences between our vector and deque | implementations, I wrote something about it in private mail to Howard, | but we didn't pursue the issue much...
I wasn't aware of that project. We now have more incentive to puruse it :-) | Coming to the specific issue, and *assuming* we agree about the | usefulness of warning for signed -> unsigned, maybe we are "lucky", in | this sense: Table 65 says *explicitely* that size_type can represent | any non-negative value of difference_type. Thus, as I see the issue, | we can safely cast from the latter to the former, if we want, without | entering in the larger discussions mentioned above... What do you | think? My concern is for vectors whose size() exceeds numeric_limits<difference_type>::max() -- even if we would like to avoid the general discussion. My understanding -- and I would like to hear of others' -- is that the semantics description in the table 65 is a constraint on implementations. Concretely, it means that we cannot have a vector whose size() is numeric_limits<size_type>::max() because otherwise the difference between end() and begin() will overflow. That implies that max_size() would have to return numeric_limit<difference_type>::max(). -- Gaby
