RFR: JDK-8304439: Subscription based listeners [v2]

[Kevin Rushforth]

On Fri, 23 Jun 2023 21:57:03 GMT, Kevin Rushforth <kcr at openjdk.org> wrote:

>> On that same topic of naming methods:
>> 
>> What do people think of `Subscription#unsubscribe`?  Should it be `cancel`? Something else?  Okay as it is?
>> 
>> Code example:
>> 
>>         if (subscription != null) {
>>             subscription.unsubscribe();
>>             subscription = null;
>>         }
>
>> | invalidation | values | changes |
>> |---|---|---|
>> |`subscribe(Runnable)`(*)|`subscribe(Consumer)`(*)|`subscribe(BiConsumer)`(*)|
>> |`subscribeInvalidations(Runnable)`|`subscribeValues(Consumer)`|`subscribeChanges(BiConsumer)`|
>> |`invalidationsTo(Runnable)`|`valuesTo(Consumer)`|`changesTo(BiConsumer)`|
>> 
>> (*) May limit future listener types that have same arity, but can still be a good choice
> 
> My preference is in the order you listed them. If we go with `subscribe`, using method name overloading, do want to add a future overload that takes a type with the same arity as one of the existing three, we can always assign a new name to that new method (since adding another overload wouldn't be source compatible, we would likely need a new name at that point).
> 
> @nlisker @andy-goryachev-oracle @mstr2 - what do you think?

> On that same topic of naming methods:
> 
> What do people think of `Subscription#unsubscribe`? Should it be `cancel`? Something else? Okay as it is?

OK as is. Especially if we go with one of the two choices with "subscribe" in the name.

-------------

PR Review Comment: https://git.openjdk.org/jfx/pull/1069#discussion_r1240441669