[Zeev Suraski <[EMAIL PROTECTED]>]
> At 13:34 16/5/2001, Stig Sæther Bakken wrote:
> >Acknowledged. But IMO the arrays will be either clearly continous or
> >clearly associative/non-continous in 99.2% of the cases (Zeev
> >statistics applied).
>
> I never claim to be accurate, I usually say 99%
At 13:34 16/5/2001, Stig Sæther Bakken wrote:
>Acknowledged. But IMO the arrays will be either clearly continous or
>clearly associative/non-continous in 99.2% of the cases (Zeev
>statistics applied).
I never claim to be accurate, I usually say 99% :)
> I think it's fine to keep treating an
Acknowledged. But IMO the arrays will be either clearly continous or
clearly associative/non-continous in 99.2% of the cases (Zeev
statistics applied). I think it's fine to keep treating an array as
non-continous if there has ever been holes or string keys in it.
- Stig
[Zeev Suraski <[EMAIL
It doesn't necessarily mean having to scrap the whole idea, just that we'd
actually have to count elements, instead of just flagging. I'll try to
think if there are any problems with this approach.
Zeev
At 02:57 15/5/2001, Andrei Zmievski wrote:
>At 02:51 AM 5/15/01 +0300, Zeev Suraski wrote:
At 02:51 AM 5/15/01 +0300, Zeev Suraski wrote:
>Ok, if we humor ourselves with this feature... What kind of behavior
>would you expect if a key gets deleted, and there are no longer
>associative members in the array?
Good point... any time savings on the extension side would be negated if
the
Ok, if we humor ourselves with this feature... What kind of behavior would
you expect if a key gets deleted, and there are no longer associative
members in the array?
Zeev
At 02:42 15/5/2001, Harald Radi wrote:
> > Any XML-RPC implementation would benefit from seeing easily whether a
> > valu
