Re: [Jmol-developers] static success, I think....

[Miguel <mi...@jm...>]

Nico wrote:
> Bob Hanson wrote:
>> pps. The following is more elegant and also works:
>>
>>   static Boolean=5B=5D testInit =3D new Boolean=5B1=5D;
>>   private void initialize() =7B
>>     if (testInit=5B0=5D =21=3D null) =7B
=5Bsnip=5D
>> It takes advantage of the fact that static testInit=5B0=5D has THREE
>> possible values: null, false, and true.
>> This is what I'm going with for now.
>>
>
> That won't work in every situation :
> - thread 1 executes =22if (testInit=5B0=5D =21=3D null)=22 and finds th=
at it evals
> to false
> - thread 2 executes =22if (testInit=5B1=5D =21=3D null)=22 and finds th=
at it evals
> to false also
> - thread 1 goes on
> - thread 2 goes on
>
> You could try using a synchronization on a variable.

Nico is correct.

It is not possible to =22roll your own=22 synchronization in a high level=

language without having some type of atomic 'test-and-set' operation
support at a lower level.

Essentially all CPUs for the last 50 years have had this, but popular
high-level languages have not exposed this functionality. Java is arguabl=
y
the first commercially-successful-and-broadly-accepted language that
incorporates synchronization primitives into the language.

Per Nico's other message, my bug with 'initialize' is that it was marked
'synchronized' instead of 'syncronized static'.


Miguel