Feedback for records: Accessors name() vs. getName()

[Brian Goetz]

I get your concern, and we'll think about what we can do.  But in the 
meantime, I think it's not quite as black-and-white as you paint it.  
Frameworks that lean on reflection generally use a bag of tricks, not 
just one trick, to try to infer design intent.  Given that records have 
a clear way to do this (Class::isRecord and 
Class::getRecordComponents**), that's one more tool for the bag, which 
can be added without dropping something else from the bag.  There may be 
other tools coming, but it's not either-or.


**You might ask why we call this `getRecordComponents` and not 
`recordComponents`; simply, because this is an exisiting, extremely 
widely used API that has already picked its conventions, and it was 
deemed better to be locally consistent in this situation.

On 8/27/2020 11:55 AM, Roman Elizarov wrote:
> Brian,
>
> thanks for the detailed answer.
>
> > As I've hinted in other places, we want to generalize this notion of 
> accessor to classes that are not records;
> (and interface, too, I suppose).
>
> And that's exactly my conundrum. I'd really like to see some guidance 
> in this direction coming directly from Java and JVM platform. It would 
> be best to see it at the same moment when records are 
> released. Otherwise, there is a big risk of fragmentation in the 
> ecosystem as different frameworks will start coming with different 
> incompatible ways to mark the method as an accessor, the role that was 
> historically a prerogative of "get" verb in the name, but now will 
> have to be done in some other ways in light of records. As you 
> correctly noted, frameworks will not have much choice but to designate 
> an annotation for this role. Even if a JVM platform just introduces an 
> annotation that does not do anything of much substance, but declares a 
> developer's intent to have an accessor method, it would still be 
> better than nothing and will help big time increasing consistency in 
> the JVM ecosystem, just like the notion of "getXxx" getter methods was 
> once brought in and codified by the JVM platform itself.
>
> Sincerely,
> Roman Elizarov