Changing array key text

[Brian Milby]

Value object starts with a ref count of 1 in the array
The first put increases the ref count to 2
The delete takes it back to 1
The second put takes it back to 2
At the end of the handler it goes back to 1 since the local var goes away

Thanks,
Brian
On Jun 25, 2019, 6:35 PM -0400, Mark Wieder via use-livecode <use-livecode at lists.runrev.com>, wrote:
> On 6/25/19 1:25 AM, Mark Waddingham via use-livecode wrote:
> > On 2019-06-25 04:18, Brian Milby via use-livecode wrote:
> > > My guess is that it has to do with the copy on write property of
> > > arrays.  If you used your method it may force an actual duplication.
> > >  The existing method just passes a pointer effectively.  I could be
> > > totally wrong though.
> > >
> > > On Jun 24, 2019, 10:11 PM -0400, Mark Wieder via use-livecode
> > > <use-livecode at lists.runrev.com>, wrote:
> > > > > put xArray[item 1 of pPath] into tSubArray
> > > > > delete variable xArray[item 1 of pPath]
> > > > > put tSubArray into xArray[pKey]
> > > >
> > > > Interesting. Wouldn't it be faster to do
> > > >
> > > > put xArray[item 1 of pPath] into xArray[pKey]
> > > > delete variable xArray[item 1 of pPath]
> > > >
> > > > i.e., why the extra copy?
> >
> > In terms of copies there are none happening in either case - only xArray
> > is being mutated - and referencing counting means that the 'puts' are
> > all free.
> >
> > The first will be 'on average' marginally more efficient but that's
> > because of the number of keys...
> >
> > In method (1) the number of keys goes from N -> N - 1 -> N
> >
> > In method (2) the number of keys goes from N -> N + 1 -> N
> >
> > Its much more likely that adding a key will cause the engine to have to
> > extend the internal vector of key-value pairs it uses to store the array
> > mapping then removing a key will cause it to shrink it - also expansion
> > is 'on average' slower than shrinking.
>
> Thanks. That helps. Mostly.
> If no copies are made then I'm still a bit weirded out by
>
> put xArray[item 1 of pPath] into tSubArray
> delete variable xArray[item 1 of pPath]
> put xArray[item 1 of pPath] into limbo
> put tSubArray into xArray[pKey]
>
> where limbo is empty and tSubArray still retains the contents.
> The quantum effect jokes write themselves.
>
> Is it the case the tSubArray retains a pointer to the contents but the
> entry in the hash is otherwise invalidated but not garbage collected
> until the reference count goes to zero?
>
> --
> Mark Wieder
> ahsoftware at gmail.com
>
> _______________________________________________
> use-livecode mailing list
> use-livecode at lists.runrev.com
> Please visit this url to subscribe, unsubscribe and manage your subscription preferences:
> http://lists.runrev.com/mailman/listinfo/use-livecode