Array: push_back on an uninitialised array fails as a side effect

[adbridge]

Currently the push_back() code looks like this:

   bool push_back(const T& new_element) {
        unsigned idx = _elements, prev_capacity = _capacity;
        if (_elements == _capacity) {
            if (_grow_capacity == 0) { // can we grow?
                return false;
            }

If this method is called on an uninitialised array then we would expect it to return False, but it does so only as a side effect of trying to test whether the array is allowed to grow. ie _elements, _capacity and _grow_capacity should all start as 0

It would be safer to have a specific test for the array not having been initialised yet.

[bogdanm]

It would be safer to have a specific test for the array not having been initialised yet.

Absolutely! Although with the kind of changes we're planning init might disappear completely, so this might get better all on its own.

[adbridge]

We may well have to fix this and the the initialisation of default values asap ready for the Q1 release as the optimisations may not get the go ahead until Q2 ?

[ciarmcom]

ARM Internal Ref: IOTSFW-2012

[0xc0170]

73 fixes this