Can QHash::capacity() be smaller than QHash::size()?!?

[iw2nhl]

I'm trying to optimize memory usage in my code.
I found that some Qt containers have capacity(), reserve() and squeeze() functions for this purpose.
Making some tests I found a strange result (among normal ones!):
in some conditions I get capacity() < size().
If I understand correctly the documentation, capacity() should give the number of allocated items (which generally is greater than the number of inserted items, for speed reasons).

QHash< DataUniqueId, DrawItem* > m_lookupTable;
 
// Insert data into m_lookupTable
for (...) {
    m_lookupTable.insert(..., ...);
}
 
qDebug() << m_lookupTable.size() << m_lookupTable.capacity();
 
// No more insertions, so release any unused memory
m_lookupTable.squeeze();
 
qDebug() << m_lookupTable.size() << m_lookupTable.capacity();

QHash< DataUniqueId, DrawItem* > m_lookupTable;

// Insert data into m_lookupTable
for (...) {
    m_lookupTable.insert(..., ...);
}

qDebug() << m_lookupTable.size() << m_lookupTable.capacity();

// No more insertions, so release any unused memory
m_lookupTable.squeeze();

qDebug() << m_lookupTable.size() << m_lookupTable.capacity();

To copy to clipboard, switch view to plain text mode 

This is the output:

500 521
500 257

500 521
500 257

To copy to clipboard, switch view to plain text mode 

As you can see, after the squeeze(), size() > capacity().
How can the number of allocated items be less than the number of real items?
Am I doing something wrong?

Thanks in advance,
Alessandro

[jacek]

If I understand correctly the documentation, capacity() should give the number of allocated items (which generally is greater than the number of inserted items, for speed reasons).

In case of QHash the docs say:

int QHash::capacity () const
Returns the number of buckets in the QHash's internal hash table.

Buckets are not items --- they might contain more than one (key, value) pair.

[iw2nhl]

Buckets are not items --- they might contain more than one (key, value) pair.

Thank you very much for the explanation!
Now I'm more secure about my code...