Hi,
I was doing a code review and noticed something that I'd never noticed
before and I wondered if it was 'legal' (aka a good idea) to do it this
way. Please let me know what you think.
We have a base class called nsXFormsAccessorsBase which implements the
scriptable interface nsIXFormsAccessors. nsXFormsAccessorsBase is a
class that isn't meant to be instantiated, but rather it is subclassed
by nsXFormsAccessors and nsXFormsControlAccessors. So the base class
provides a lot of the functionality for the methods specified in the
interface, but the subclasses override a couple of the functions to get
the data to return from different places.
So for example, a form author in JS can call var accessors =
myfoo.getAccessorsByNode(aNode). Internally this does:
Foo::GetAccessorsByNode(nsIDomNode *aNode, nsIXFormsAccessors **aAccessors)
{
...
*aAccessors = new nsXFormsAccessors(this, aNode);
if (!*aAccessors)
return NS_ERROR_OUT_OF_MEMORY;
NS_ADDREF(*aAccessors);
return NS_OK;
}
I've never seen code like this in Mozilla, where someone takes the
result of a 'new', addrefs it and returns it directly to the calling
function (whether it be in c++, JS, etc.). I've debugged this function
being called from JS and I see the memory being freed up during the JS
garbage collection, so it seems to work and have the proper lifespan and
all.
Could someone please verify for me that this is a proper way to do
something like this?
Thanks,
--Aaron
_______________________________________________
dev-tech-xpcom mailing list
[email protected]
https://lists.mozilla.org/listinfo/dev-tech-xpcom
