Re: Basic question on weak keys

[Andrew Starks <andrew.starks@...>]

On Jun 13, 2012, at 17:17, Rena <hyperhacker@gmail.com> wrote:

> On Wed, Jun 13, 2012 at 4:02 PM, Peter Cawley <lua@corsix.org> wrote:
>> On Wed, Jun 13, 2012 at 10:59 PM, Andrew Starks <andrew.starks@trms.com> wrote:
>>> b = nil
>>> print(a[b])
>>> --> nil
>>
>> As `b` is now nil, `a[b]` means `a[nil]`, and as `nil` is never a
>> valid table key, `a[nil]` will give `nil` for all tables `a`.
>>
>
> Your check is broken: you set b=nil, so now you're looking up a[nil],
> not a[original value of b].
>
> Using an object to index a table doesn't keep any reference to it.
> Only storing it in the table, as a key. Weak tables tell the garbage
> collector to ignore the keys/values of this table when checking for
> references. So, if you didn't have a reference to that key/value
> anywhere else, it will eventually be collected and the pair removed
> from the table. (If the key is removed, the value is also removed, and
> vice-versa, since a key without a value is useless, and a value
> without a key impossible.)
>
> --
> Sent from my Game Boy.
>

Awesome. I like to think I'm smart about references. Clearly I'm not.

So, if I looped with pairs(a), it would be there. __mode = 'k' and it
would not. Correct?