Allocating vector of strings seem to crash. Allocating array ofstrings seems to be ok .

R

Rakesh Kumar

Hi All -
In a project of mine - I was trying to scale down the actual issue
to the following piece of code. I need to allocate an array of strings
and reserve the individual string to a particular size (4K) .
I wrote 2 functions - allocVectorOfStrings() and
allocArrayOfStrings().
Each of them seem to allocate similar amounts of memory - but the
version of vectorOfStrings seem to crash with the following error -
"double free or corruption (out): 0x08055ff8 ***" .

I was just curious if I am doing anything fundamentally wrong here
to cause the issue.


#include <iostream>
#include <cstdlib>
#include <vector>


void allocVectorOfStrings();
void allocArrayOfStrings();

void allocVectorOfStrings() {
std::vector<std::string> * vt = new std::vector<std::string>();
vt->reserve(50);
for (size_t i = 0; i < vt->capacity(); ++i) {
std::cout << "We are probably ok" << i << "\n";
vt->operator[](i).reserve(40);
}
delete vt;
}

void allocArrayOfStrings() {
std::string * vt = new std::string[4096];
for (size_t i = 0; i < 1024; ++i) {
std::cout << "We are probably ok" << i << "\n";
vt.reserve(4096);
}
delete [] vt;
}

int main()
{
allocArrayOfStrings();
allocVectorOfStrings(); //This function crashes
return EXIT_SUCCESS;
}
 
V

Victor Bazarov

Rakesh said:
Hi All -
In a project of mine - I was trying to scale down the actual issue
to the following piece of code. I need to allocate an array of strings
and reserve the individual string to a particular size (4K) .
I wrote 2 functions - allocVectorOfStrings() and
allocArrayOfStrings().
Each of them seem to allocate similar amounts of memory - but the
version of vectorOfStrings seem to crash with the following error -
"double free or corruption (out): 0x08055ff8 ***" .

I was just curious if I am doing anything fundamentally wrong here
to cause the issue.


#include <iostream>
#include <cstdlib>
#include <vector>


void allocVectorOfStrings();
void allocArrayOfStrings();

void allocVectorOfStrings() {
std::vector<std::string> * vt = new std::vector<std::string>();
vt->reserve(50);

'reserve' does not construct vector's elements. It only allocates
memory for constructing them later, when 'insert' is used. Here
'*vt' does not contain _any strings_. It's empty. It's _capable_
of containing at least 50 without reallocation of its storage. But
it doesn't have any elements.
for (size_t i = 0; i < vt->capacity(); ++i) {

This is a very bad idea. Never iterate to capacity. Always
iterate to size.
std::cout << "We are probably ok" << i << "\n";

No, you're not OK.
vt->operator[](i).reserve(40);

You're accessing a non-existent element at the index 'i' and
then calls a member functions for it. Undefined behaviour.
}
delete vt;
}

void allocArrayOfStrings() {
std::string * vt = new std::string[4096];

Here 'vt' contains all 4096 elements.
for (size_t i = 0; i < 1024; ++i) {
std::cout << "We are probably ok" << i << "\n";

Yes, you are.
vt.reserve(4096);


'vt' represents a real string. You can call 'reserve' for
it, no problem.
}
delete [] vt;
}

int main()
{
allocArrayOfStrings();
allocVectorOfStrings(); //This function crashes
return EXIT_SUCCESS;
}

V
 
S

Salt_Peter

Hi All -
In a project of mine - I was trying to scale down the actual issue
to the following piece of code. I need to allocate an array of strings
and reserve the individual string to a particular size (4K) .
I wrote 2 functions - allocVectorOfStrings() and
allocArrayOfStrings().
Each of them seem to allocate similar amounts of memory - but the
version of vectorOfStrings seem to crash with the following error -
"double free or corruption (out): 0x08055ff8 ***" .

I was just curious if I am doing anything fundamentally wrong here
to cause the issue.

#include <iostream>
#include <cstdlib>
#include <vector>

void allocVectorOfStrings();
void allocArrayOfStrings();

void allocVectorOfStrings() {
std::vector<std::string> * vt = new std::vector<std::string>();
vt->reserve(50);
for (size_t i = 0; i < vt->capacity(); ++i) {
std::cout << "We are probably ok" << i << "\n";
vt->operator[](i).reserve(40);

Try not to use new / delete unless you are required to do so.
You are attempting to access an element that doesn't yet exist.
reserve constructs nothing.
use vector's member function at(...) when in doubt. see below.
}
delete vt;

}

void allocArrayOfStrings() {
std::string * vt = new std::string[4096];
for (size_t i = 0; i < 1024; ++i) {
std::cout << "We are probably ok" << i << "\n";
vt.reserve(4096);
}
delete [] vt;

}

int main()
{
allocArrayOfStrings();
allocVectorOfStrings(); //This function crashes
return EXIT_SUCCESS;

}


#include <iostream>
#include <string>
#include <vector>
#include <stdexcept>

void allocVectorOfStrings()
{
std::vector<std::string> vt(10, "default string");
std::cout << "vt's size is " << vt.size() << std::endl;
for(size_t i = 0; i < vt.size(); ++i)
{
std::cout << "vt[ " << i << " ] ";
std::cout << vt.at(i) << std::endl;
}
// uncomment for a test
// vt.at(10) = "out or range";
}

int main()
{
try
{
allocVectorOfStrings();
}
catch( const std::exception& e)
{
std::cout << "Error:";
std::cout << e.what() << std::endl;
}
}

/*
vt's size is 10
vt[ 0 ] default string
....
vt[ 9 ] default string
*/
 
R

Rakesh Kumar

Rakesh said:
Hi All -
In a project of mine - I was trying to scale down the actual issue
to the following piece of code. I need to allocate an array of strings
and reserve the individual string to a particular size (4K) .
I wrote 2 functions - allocVectorOfStrings() and
allocArrayOfStrings().
Each of them seem to allocate similar amounts of memory - but the
version of vectorOfStrings seem to crash with the following error -
"double free or corruption (out): 0x08055ff8 ***" .
I was just curious if I am doing anything fundamentally wrong here
to cause the issue.
#include <iostream>
#include <cstdlib>
#include <vector>
void allocVectorOfStrings();
void allocArrayOfStrings();
void allocVectorOfStrings() {
std::vector<std::string> * vt = new std::vector<std::string>();
vt->reserve(50);

'reserve' does not construct vector's elements. It only allocates
memory for constructing them later, when 'insert' is used. Here
'*vt' does not contain _any strings_. It's empty. It's _capable_
of containing at least 50 without reallocation of its storage. But
it doesn't have any elements.
for (size_t i = 0; i < vt->capacity(); ++i) {

This is a very bad idea. Never iterate to capacity. Always
iterate to size.
std::cout << "We are probably ok" << i << "\n";

No, you're not OK.
vt->operator[](i).reserve(40);

You're accessing a non-existent element at the index 'i' and
then calls a member functions for it. Undefined behaviour.

Thanks Victor.
The revised function seems to do what I intended in the first place.

void allocVectorOfStrings()
{
std::vector<std::string> vt(1024);

for (size_t i = 0; i < vt.size(); ++i)
{
std::cout << "Vector seems to be ok too" << i << "\n";
vt.reserve(4096);
}
}

Just a quick question - after a vector is allocated - is there a way I
can mass construct elements at one shot (instead of using insert /
push_back ) - similar to the construct shown above.



}
delete vt;
}
void allocArrayOfStrings() {
std::string * vt = new std::string[4096];

Here 'vt' contains all 4096 elements.
for (size_t i = 0; i < 1024; ++i) {
std::cout << "We are probably ok" << i << "\n";

Yes, you are.
vt.reserve(4096);


'vt' represents a real string. You can call 'reserve' for
it, no problem.
}
delete [] vt;
}
int main()
{
allocArrayOfStrings();
allocVectorOfStrings(); //This function crashes
return EXIT_SUCCESS;
}

V
 
W

werasm

'reserve' does not construct vector's elements. It only allocates
memory for constructing them later, when 'insert' is used. Here
'*vt' does not contain _any strings_. It's empty. It's _capable_
of containing at least 50 without reallocation of its storage. But
it doesn't have any elements.
This is a very bad idea. Never iterate to capacity. Always
iterate to size.
No, you're not OK.
vt->operator[](i).reserve(40);

You're accessing a non-existent element at the index 'i' and
then calls a member functions for it. Undefined behaviour.

Thanks Victor.
The revised function seems to do what I intended in the first place.

void allocVectorOfStrings()
{
std::vector<std::string> vt(1024);

for (size_t i = 0; i < vt.size(); ++i)
{
std::cout << "Vector seems to be ok too" << i << "\n";
vt.reserve(4096);
}

}

Just a quick question - after a vector is allocated - is there a way I
can mass construct elements at one shot (instead of using insert /
push_back ) - similar to the construct shown above.


vt.resize( x );

After resize is used the vector will contain x elements. You could
also re-assign to it or use swap e.g:

vt = std::vector<std::string>( x ); //or

std::vector<std::string> v( x )
vt.swap( v );

I think resize would most probably be the best option though.

Regards,

Werner
 
J

James Kanze

Rakesh said:
In a project of mine - I was trying to scale down the actual issue
to the following piece of code. I need to allocate an array of strings
and reserve the individual string to a particular size (4K) .
I wrote 2 functions - allocVectorOfStrings() and
allocArrayOfStrings().
Each of them seem to allocate similar amounts of memory - but the
version of vectorOfStrings seem to crash with the following error -
"double free or corruption (out): 0x08055ff8 ***" .
I was just curious if I am doing anything fundamentally wrong here
to cause the issue.
#include <iostream>
#include <cstdlib>
#include <vector>
void allocVectorOfStrings();
void allocArrayOfStrings();
void allocVectorOfStrings() {
std::vector<std::string> * vt = new std::vector<std::string>();
vt->reserve(50);
'reserve' does not construct vector's elements. It only allocates
memory for constructing them later, when 'insert' is used. Here
'*vt' does not contain _any strings_. It's empty. It's _capable_
of containing at least 50 without reallocation of its storage. But
it doesn't have any elements.
for (size_t i = 0; i < vt->capacity(); ++i) {
This is a very bad idea. Never iterate to capacity. Always
iterate to size.
std::cout << "We are probably ok" << i << "\n";
No, you're not OK.
vt->operator[](i).reserve(40);
You're accessing a non-existent element at the index 'i' and
then calls a member functions for it. Undefined behaviour.

Still, a good implementation of the library will output a more
or less sensible error message when you do it. I get:
error: attempt to subscript container with out-of-bounds index 0,
but
container only holds 0 elements.
when I try it with g++, for example (provided I've activated
debugging, but I always do). And VC++ indicates "Expression:
vector subscript out of range" in its pop-up box. Rather clear
as well.
Thanks Victor.
The revised function seems to do what I intended in the first place.
void allocVectorOfStrings()
{
std::vector<std::string> vt(1024);
for (size_t i = 0; i < vt.size(); ++i)
{
std::cout << "Vector seems to be ok too" << i << "\n";
vt.reserve(4096);
}
}

Just a quick question - after a vector is allocated - is there
a way I can mass construct elements at one shot (instead of
using insert / push_back ) - similar to the construct shown
above.

In the above, the elements are mass constructed, in the
constructor of vt. In this case, they are copy constructed from
the default, std::string(), but you can pass any initial value
you want, and they will be copy constructed from that.

On the other hand, the capacity() is not part of the "value" of
a string, and is not necessarily copied by the copy constructor;
in order to ensure the capacity of all of the elements, I think
you do have to do something like the above. Alternatively, you
could do:

std::vector< std::string > vt(
1024,
std::string( 4096, '\0') ) ;

, which will ensure the actual size (and not just the capacity).
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Members online

No members online now.

Forum statistics

Threads
473,995
Messages
2,570,228
Members
46,818
Latest member
SapanaCarpetStudio

Latest Threads

Top