Access violation using vector::reserve()

[mphil1974]

Hello,

I don't understand how the problem I'm having is even possible--

lines.reserve(3 ); //Access violation

'lines' is a vector of structs ('Line'). It doesn't matter whether I use a vector<Line>::s ize_type object in place of '3', I get the same result.

Crucially, 'lines' is *not* empty at this point--the first element has been assigned.

The books I've looked at (e.g. Josuttis) have given me no indication that a post-assignment call to 'reserve' should be illegal--is it?

I'm new to C++ (but not C, Matlab), and I have run into errors arising from sloppiness on my part with 'const'--could that have something to do with it?

Again, if people have ideas as to how this error is even possible I would appreciate it (posting all my code seems like it would just cloud the issue at this point).

Thanks--
Best,
Matt

[boxfish]

Hi,
Did your program first get ahold of that vector by calling a function that you defined? If it did, maybe you forgot the return statement on that function. I've gotten an access violation before from a missing a return statement. Hope this helps.

[mphil1974]

No, I think my compiler (DevC++ 4.9.9.2) catches all of those (in any case, didn't happen this time).

I should perhaps have been more clear--the access violation happens literally at the point of the reserve statement--not one statement before or one statement after.

Best,
Matt

[Laharl]

Based on the docs I've seen, it's not the assignment. Declaring it as const, however, probably would make reserve() fail.

[weaknessforcats]

Soemthing else is going on. This code compiles and runs OK using Visual Studio.NET 2008:

[code=cpp]
struct Line
{

};
int main()
{
vector<Line> lines;
lines.reserve(3 ); //OK}
[/code]

[boxfish]

When I use reserve on a vector that was declared as const, I get some kind of compile error, not an access violation. I'm using Dev-C++ 4.9.8.0. The only thing I can think of is that there's something wrong with your vector. Are you sure posting some more of your code wouldn't help?

[weaknessforcats]

When you create a const, you can't change it. The reserve() would change the vector and that will cause a compiler error.

[mphil1974]

Hello,

Thank you everybody for your input, I'm glad to hear that const issues can be ruled out as the program compiled just fine.

In fact it now looks like I wasted everyone's time, as I was able to trace the error back through STL to a copy constructor and an attempt to access info that didn't exist. The Line struct contained a member which is a recursive object (I'm writing a natural deduction program) so I had to create my own ctors. I still maintain not subjecting you to my code was the right thing to do and you would too if you saw it :).

A related issue which I found even stranger, and which also went away when I resolved the ctor issue, was

cout << lines.capacity( ); //Access violation if this line NOT included!!

That's right--the program crashed *unless* this call to lines.capacity was made, and the result outputted. Was this some kind of DevC++ quirk? What does vector::capacit y do?

Best,
Matt

[weaknessforcats]

vector::capacit y() returns the current number of objects the vector can contain without allocating more memory.

[mphil1974]

vector::capacit y() returns the current number of objects the vector can contain without allocating more memory.

Right--I meant, what if any calls to possibly user-defined functions does vector::capacit y make? Does it call ctors or overloadable operators like '==' ? Thanks--

Best,
Matt

[weaknessforcats]

vector::capacit y() is a const member funciton so it cannot change any data in the vector. That means it cannot call any user-defined functions.

[mphil1974]

vector::capacit y() is a const member funciton so it cannot change any data in the vector. That means it cannot call any user-defined functions.

Okay, thank you!

Best,
Matt