«

Hi all,

I've got a simple question regarding stl containers. Consider this
code:

std::vector<float> foo;
foo.resize(100);

How do I know if memory allocation was successful? The resize method
returns nothing. I understand that resize does not always allocate
memory. But even if I use the reserve method it doesn't return boolean
or something to validate.

Should I do something like this:

if(foo.capacity()<foo.size())
cout << "Memory allocation fail!" << endl;

Is there any standard way of doing this.

Thanks all
Kwijibo

"kwijibo28" <> wrote...
> I've got a simple question regarding stl containers. Consider this
> code:
>
> std::vector<float> foo;
> foo.resize(100);
>
> How do I know if memory allocation was successful? The resize method
> returns nothing. I understand that resize does not always allocate
> memory.

If it does, and there isn't enough of it, std::bad_alloc is thrown.

> But even if I use the reserve method it doesn't return boolean
> or something to validate.
>
> Should I do something like this:
>
> if(foo.capacity()<foo.size())
> cout << "Memory allocation fail!" << endl;
>
> Is there any standard way of doing this.

Yes,

try {
foo.resize(100);
}
catch (std::bad_alloc const&) {
cout << "Memory allocation fail!" << endl;
}

Victor

"kwijibo28" wrote:
> Hi all,
>
> I've got a simple question regarding stl containers. Consider this
> code:
>
> std::vector<float> foo;
> foo.resize(100);
>
> How do I know if memory allocation was successful? The resize method
> returns nothing. I understand that resize does not always allocate
> memory. But even if I use the reserve method it doesn't return boolean
> or something to validate.

Memory allocation requests don't fail. They succeed or throw an exception
(std::bad_alloc).

HTH,
Patrick

kwijibo28 wrote:
> Hi all,
>
> I've got a simple question regarding stl containers. Consider this
> code:
>
> std::vector<float> foo;
> foo.resize(100);
>
> How do I know if memory allocation was successful? The resize method
> returns nothing. I understand that resize does not always allocate
> memory. But even if I use the reserve method it doesn't return boolean
> or something to validate.
>
> Should I do something like this:
>
> if(foo.capacity()<foo.size())
> cout << "Memory allocation fail!" << endl;
>
> Is there any standard way of doing this.

Research the 'std::bad_alloc' exception class. The implementation (i.e.,
the standard C++ library) throws a std::bad_alloc object to report
storage allocation failures.

<example>
try {
std::vector<float> foo;
...
foo.resize(100);
...
}
catch ( const std::bad_alloc & error) {
// whoops... if control reaches here, a memory allocation
// failure occurred somewhere within the 'try{}' block
}
</example>

--
Jim

To reply by email, remove "link" and change "now.here" to "yahoo"
jfischer_link5809{at}now.here.com

kwijibo28 wrote:
>
> Thanks for the reply guys.
> I've tried all this but it still doesn't work. So I'll be a little
> more specific about what I am trying do. I'm writing an MFC aplication
> using VC++ 6.0.
>
> When I write something like:
>
> try {
> foo.resize(power(2,32)-1);
> }
> catch (std::bad_alloc const&) {
> ::MessageBox(NULL,"Memory allocation fail!","Error",MB_OK);
> }
>
> I've got a message box that pop up with a "Out of memory" message when
> foo.resize is called. I don't know where this message box come from.

Deep from the internals of your runtime system.
MS has yet to learn that emitting such messages is just a way to
support lazy programmers. The number of programmers trying to get
rid of the floppy-disk-failed message in DOS is legion. MS has not
learned from that.

> And even worst, my message box with "Memory allocation fail!" don't
> appear.

VC++ 6.0 does not throw an exception when a memory allocation failes.
Instead NULL is returned from new. Don't know what resize does with
that information.

But honestly: 2 to the power of 32 is a lot of memory. If we assume
that foo holds unsigned characters, that would be 4 GB. Tracing your
post back, I see that foo holds floats, so that would sum up to
16 GB in VC++. That's a lot. Are you sure you need that much memory?

--
Karl Heinz Buchegger

Karl Heinz Buchegger <> wrote in message
> >
> > When I write something like:
> >
> > try {
> > foo.resize(power(2,32)-1);
> > }
> > catch (std::bad_alloc const&) {
> > ::MessageBox(NULL,"Memory allocation fail!","Error",MB_OK);
> > }
> >
> But honestly: 2 to the power of 32 is a lot of memory. If we assume
> that foo holds unsigned characters, that would be 4 GB. Tracing your
> post back, I see that foo holds floats, so that would sum up to
> 16 GB in VC++. That's a lot. Are you sure you need that much memory?

I usually don't need that much memory but that was just a test to be
sure that memory allocation would fail, because I was trying to catch
the exception. I don't think it's good practice not to check for
memory allocation problem just because it's a small array.

kwijibo28

On 11 Aug 2003 04:14:09 -0700, (kwijibo2
wrote:

>Karl Heinz Buchegger <> wrote in message
>> >
>> > When I write something like:
>> >
>> > try {
>> > foo.resize(power(2,32)-1);
>> > }
>> > catch (std::bad_alloc const&) {
>> > ::MessageBox(NULL,"Memory allocation fail!","Error",MB_OK);
>> > }
>> >
>> But honestly: 2 to the power of 32 is a lot of memory. If we assume
>> that foo holds unsigned characters, that would be 4 GB. Tracing your
>> post back, I see that foo holds floats, so that would sum up to
>> 16 GB in VC++. That's a lot. Are you sure you need that much memory?
>
>I usually don't need that much memory but that was just a test to be
>sure that memory allocation would fail, because I was trying to catch
>the exception. I don't think it's good practice not to check for
>memory allocation problem just because it's a small array.

In your case you've passed such a huge number that it will throw a
std::length_error, since the memory allocator doesn't support
allocating that much memory even if your computer has it physically.

A better test would be:

try
{
while(1)
{
int* i = new int[10000000];
//leak!
}
}
catch(std::bad_alloc const&)
{
}

but VC6 won't throw here anyway since it doesn't follow the standard
in this respect, but VC7.1 (and maybe 7.0) catches the exception.

Tom