[Vm-dev] [Pharo-dev] About ~= and ~~

[stepharo]

On Sat, 26 Nov 2016 17:51:24 +0100, Eliot Miranda  
<eliot.miranda at gmail.com> wrote:

> Hi Levente,
>
>
>> On Nov 26, 2016, at 3:02 AM, Levente Uzonyi <leves at caesar.elte.hu>  
>> wrote:
>>
>>> On Sat, 26 Nov 2016, Clément Bera wrote:
>>>
>>> If the send is to a non failing primitive call, there is no suspension
>> point either. So I don't think it would break programs.
>>
>> When there's a send, one can override the method, so the guaratee of no  
>> suspension points is gone.
>
> Right, but as the doctor says, "don't do that". For example, provided  
> that the method is only overridden in exotic classes such as proxy  
> classes, and is not overridden for nil and things like LinkedList,  
> Semaphore and Process then it can still be used without fear of  
> suspension points in the scheduler.
>
> And we can think of adding better concurrency control primitives such as  
> a conditional assignment operator.

Yes I would love that.
It would be so great to have an **EXPLICIT** constructs. It would improve  
the code quality
because relying on the underlying assumptions of the VM implementation  
choice is a lack
of language design in fact.

Stef


>
>> Levente


-- 
Using Opera's mail client: http://www.opera.com/mail/