iterator_adapter conversion to base iterator?
Hello everybody! Please don't blame for beeing foolish - I'm using the iterator-adapter class for the first time. I converted a standard iterator from a std::vector with the iterator_adapter stuff into another class random_set_iterator. But at certain points in my programm I need the random_set_iterator converted back to the original iterator class from the vector class. I noticed the base()-function which basically does the job, but I think there should be a more elegant and clearer way to do it. Am I wrong or shouldn't an adapted iterator almost always be convertible to it's base iterator? I'm sure, it should not hard to implement somthing like this, but how? Thanks in advance! Greetings, Sebastian
Sebastian Weber
Hello everybody!
Please don't blame for beeing foolish - I'm using the iterator-adapter class for the first time. I converted a standard iterator from a std::vector with the iterator_adapter stuff into another class random_set_iterator. But at certain points in my programm I need the random_set_iterator converted back to the original iterator class from the vector class. I noticed the base()-function which basically does the job, but I think there should be a more elegant and clearer way to do it. Am I wrong or shouldn't an adapted iterator almost always be convertible to it's base iterator?
It shouldn't. Implicit conversions are generally dangerous; there's no good reason to allow one here.
I'm sure, it should not hard to implement somthing like this, but how?
You can easily add an operator base_iterator_type() const to random_set_iterator, but we (the authors of iterator_adaptor) don't recommend it. -- Dave Abrahams Boost Consulting www.boost-consulting.com
Hi!
It shouldn't. Implicit conversions are generally dangerous; there's no good reason to allow one here.
Well, that might be in general the case. But at least in my case I think it does make sense: random_set_iterator is basically a std::vector<>::iterator with the noteable difference that the next-operation ++ is replaced by a function which moves the iterator to a random element in within the vector. And as I do want to delete elements from the vector which I find by the random-iterator, I need the conversion.
I'm sure, it should not hard to implement somthing like this, but how?
You can easily add an
operator base_iterator_type() const
to random_set_iterator, but we (the authors of iterator_adaptor) don't recommend it.
Thanks, that did it. Greetings, Sebastian Weber
Sebastian Weber
Hi!
It shouldn't. Implicit conversions are generally dangerous; there's no good reason to allow one here.
Well, that might be in general the case. But at least in my case I think it does make sense: random_set_iterator is basically a std::vector<>::iterator with the noteable difference that the next-operation ++ is replaced by a function which moves the iterator to a random element in within the vector. And as I do want to delete elements from the vector which I find by the random-iterator, I need the conversion.
Surely you don't *need* it. v.erase(x.base()) will work just fine.
I'm sure, it should not hard to implement somthing like this, but how?
You can easily add an
operator base_iterator_type() const
to random_set_iterator, but we (the authors of iterator_adaptor) don't recommend it.
Thanks, that did it.
You're welcome. But just remember, we warned you. Don't be surprised if you find that you are seeing unexpected effects due to the implicit conversion. For example, your iterators random_set_iterator is probably not random access, but I bet you can now write i1 < i2. Stranger effects where code compiles but does something you didn't intend are possible. Good Luck, -- Dave Abrahams Boost Consulting www.boost-consulting.com
participants (2)
-
David Abrahams -
Sebastian Weber