I need to stop using this book....template error

here's my code

Code:

1 //sum all elements at v[2*i] 2 //test on ints and doubles 3 4 #include <iostream> 5 #include <vector> 6 #include <cstdlib> 7 #include <ctime> 8 9 using namespace std; 10 11 template <typename foo_t> 12 foo_t sum_elements(vector<foo_t>& the_array){ 13 14 foo_t sum = 0; 15 vector<foo_t>::iterator x; 16 17 for(x = the_array.begin(); x != the_array.end(); x+=2) 18 sum+=*x; 19 20 return sum; 21 } 22 23 int main(void){ 24 25 const int vector_size = 20; 26 const int rand_range = 1001; 27 srand(time(nullptr)); 28 29 vector<int> v1 = {}; 30 vector<double> v2 = {}; 31 vector<int>::iterator v1_iter; 32 vector<double>::iterator v2_iter; 33 34 //populate vectors with data 35 int i = 0; 36 for(v1_iter = v1.begin(); v1_iter != v1.end(); ++v1_iter){ 37 int j = rand() % rand_range; 38 v1.push_back(j); 39 cout << "vector<int>["<<i<<"] = " << j << "\n"; 40 i++; 41 } 42 43 for(v2_iter = v2.begin(); v2_iter !=v2.end(); ++v2_iter){ 44 int i = rand() % rand_range; 45 v2.push_back(i); 46 } 47 48 //operate on data...sum elements...output results 49 cout << "the sum of vector<int> is " << sum_elements(v1) << endl; 50 cout << "the sum of vector<double> is " << sum_elements(v2) << endl; 51 52 return 0; 53 }

here's my compiler error

Code:

p345no3.cpp: In function ‘foo_t sum_elements(std::vector<foo_t>&)’: p345no3.cpp:15:4: error: need ‘typename’ before ‘std::vector<foo_t>::iterator’ because ‘std::vector<foo_t>’ is a dependent scope p345no3.cpp:15:28: error: expected ‘;’ before ‘x’ p345no3.cpp:17:8: error: ‘x’ was not declared in this scope p345no3.cpp: In instantiation of ‘foo_t sum_elements(std::vector<foo_t>&) [with foo_t = int]’: p345no3.cpp:49:59: required from here p345no3.cpp:15:4: error: dependent-name ‘std::vector<foo_t>::iterator’ is parsed as a non-type, but instantiation yields a type p345no3.cpp:15:4: note: say ‘typename std::vector<foo_t>::iterator’ if a type is meant p345no3.cpp: In instantiation of ‘foo_t sum_elements(std::vector<foo_t>&) [with foo_t = double]’: p345no3.cpp:50:62: required from here p345no3.cpp:15:4: error: dependent-name ‘std::vector<foo_t>::iterator’ is parsed as a non-type, but instantiation yields a type p345no3.cpp:15:4: note: say ‘typename std::vector<foo_t>::iterator’ if a type is meant

now I made the necessary changes and incude the typename keyword but my question is why exactly is it needed. what does that mean std::vector<foo_t> is a dependant scope. I thought the typename syntax before the function definition took care of that and allowed the type instantiation at compile time.

also for some reason my vectors aren't initializing with the random values.

When parsing the code, the compiler needs to know if vector<T>::iterator is a type or a constant, or a function. At that point in time, the compiler cannot say what it is because T is unknown. Hence, iterator is dependent upon the type T.
It cannot look into the definition of std::vector<T>, because there may be specializations of std::vector where iterator may not be a type.

Also, didn't we have this discussion before (before I derailed the thread :P)?

Change:

vector<foo_t>::iterator x;
for(x = the_array.begin(); x != the_array.end(); x+=2)

to:

for(auto x = the_array.begin(); x != the_array.end(); x+=2)

You are using gcc, so just use -std=c++11.

Also see c++ - When is the "typename" keyword necessary? - Stack Overflow.

Quote:

for(auto x = the_array.begin(); x != the_array.end(); x+=2)

O_o

It would be a lot better to use iterators directly.

You can always add the convenience function later.

Soma

Code:

template < typename FIterator > FIterator advance_iterator ( FIterator fOrigin , FIterator fTerminus , size_t fDistance ); // ... template < typename FIterator , typename FElement > FElement sum_elements ( FIterator fOrigin , FIterator fTerminus , FElement fValue , size_t fSpan = 1 ); // ...

Nice! thanks that explained it very well.

lol, thanks. I'll try to remember to use some of the new C++11 features. I've changed the code below...you should like the new brace initializer syntax :) still having issues with the vectors being initialized with their initial values though. Any thoughts ?

Code:

1 //sum all elements at v[2*i] 2 //test on ints and doubles 3 4 #include <iostream> 5 #include <vector> 6 #include <cstdlib> 7 #include <ctime> 8 9 const int vector_size = 20; 10 const int rand_range = 1001; 11 12 using namespace std; 13 14 template <typename foo_t> 15 foo_t sum_elements(vector<foo_t>& the_array){ 16 17 foo_t sum = 0; 18 typename vector<foo_t>::iterator x; 19 20 for(x = the_array.begin(); x != the_array.end(); x+=2) 21 sum+=*x; 22 23 return sum; 24 } 25 26 template<typename T> 27 void init_container(vector<T>& v){ 28 29 int i = 0; 30 typename vector<T>::iterator ndx; 31 32 for(ndx = v.begin(); ndx != v.end(); ++ndx){ 33 int j = (rand() % rand_range); 34 v.push_back(j); 35 cout << "vector<int>["<<i<<"] = " << j << "\n"; 36 i++; 37 } 38 } 39 40 int main(void){ 41 42 srand(time(nullptr)); 43 44 vector<int> v1 = {}; 45 vector<double> v2 = {}; 46 47 //capture data...init with data 48 init_container(v1); 49 init_container(v2); 50 51 //operate on data...sum elements...output results 52 cout << "the sum of vector<int> is " << sum_elements(v1) << endl; 53 cout << "the sum of vecotr<double> is " << sum_elements(v2) << endl; 54 55 return 0; 56 }

Could you get rid of the line numbers in your code? It makes copying and pasting a chore.
Also, init_container is iterating over the containers, but they are empty, so the loop never executes.
You can replace sum_elements(v1) with std::accumulate(v1.begin(), v1.end(), 0). Same for v2.

Quote:

Originally Posted by Elysia

Could you get rid of the line numbers in your code? It makes copying and pasting a chore.

Sorry, i copied it straight from vim with the :set numbers on. I'll remember that next time

Quote:

Also, init_container is iterating over the containers, but they are empty, so the loop never executes.
You can replace sum_elements(v1) with std::accumulate(v1.begin(), v1.end(), 0). Same for v2.

I need to sum the even indices in the containers, thus my own function. unless std::accumulate() does that...

You underestimate the standard library.

Code:

template<typename T> T sum_evens(T a, T b) { static bool even = false; T ret = even ? a + b : a; even = !even; return ret; } std::accumulate(v1.begin(), v1.end(), &sum_evens<int>); std::accumulate(v2.begin(), v2.end(), &sum_evens<double>);

(Not tested.)

I do not check for even numbers. I check for even invocations, which would be every 2 numbers.

Quote:

for(x = the_array.begin(); x != the_array.end(); x+=2)

O_o

This isn't safe as you could bypass the end of the array and just keep going forever.

Quote:

I need to sum the even indices in the containers, thus my own function.

Your code uses only the odd indices.

Quote:

You underestimate the standard library.

Well, that is probably true, but your code is bad and wrong for several different reasons.

1): Wrong form of `std::accumulate'.
2): Parameters passed by value.
3): Function returns the wrong value.
4): Function returns different results when passed the same data.

Soma

Quote:

Originally Posted by phantomotap

O_o

This isn't safe as you could bypass the end of the array and just keep going forever.

hmmm....yeah that would suck...much like the book I'm learning from C++ By Dissection: Ira Pohl: 9780201787337: Amazon.com: Books
don't buy it!

Quote:

Your code uses only the odd indices.

okay, how is that? if x = 0, then it will move up to x + 2 = 2, then x+2 = 4...right?

thanks,

Quote:

okay, how is that? if x = 0, then it will move up to x + 2 = 2, then x+2 = 4...right?

*derp*

You are right.

I misread your code.

Sorry for the confusion.

Soma

Quote:

Originally Posted by phantomotap

*derp*

You are right.

I misread your code.

Sorry for the confusion.

Soma

yeah no worries :) any ideas why my init_container function isn't working? I thought you could declare and empty vector and it would expand dynamically as I push_back()

O_o

Elysia has already explained the problem with `init_container' in post five.

If you don't understand that explanation, try and reason it out.

If you still don't understand post back here with what you think is right and what you think is wrong.

Soma

Quote:

Originally Posted by hex_dump

hmmm....yeah that would suck...much like the book I'm learning from C++ By Dissection: Ira Pohl: 9780201787337: Amazon.com: Books
don't buy it!

The fix is easy though, just make sure you're before the end: i < v.end()

Quote:

yeah no worries any ideas why my init_container function isn't working? I thought you could declare and empty vector and it would expand dynamically as I push_back()

It does, but you're looping over the vector, which is empty when you call that function. The loop behaves accordingly. Generating numbers has nothing to do necessarily with the size of the container, unless you preallocate.

Quote:

Originally Posted by hex_dump

any ideas why my init_container function isn't working?

Your code iterates over an empty vector;
if it were not empty you'd have an infinite loop;

Kurt

thanks guys, loops like a still have a way to go with this.

Quote:

Originally Posted by hex_dump

thanks guys, loops like a still have a way to go with this.

That's what we're here "for()" :p