[External] : Gatherer and primitive specialization

[Paul Sandoz]

> On Jun 13, 2024, at 8:20 AM, forax at univ-mlv.fr wrote:
> 
> 
> 
> From: "Viktor Klang" <viktor.klang at oracle.com>
> To: "Remi Forax" <forax at univ-mlv.fr>, "core-libs-dev" <core-libs-dev at openjdk.java.net>
> Sent: Thursday, June 13, 2024 12:52:03 PM
> Subject: Re: [External] : Gatherer and primitive specialization
> Hi Rémi,
> 
> Given that Collector has not been specialized since it was introduced, we opted to not specialize Gatherer eagerly as Valhalla Value Classes may also move the needle a bit regarding the need for specialization in general.
> 
> As i said previously, most collectors uses collections so until collections are specialized, specializing the collector API is not that useful.
> A gatherer unlike a Collector can just transform values, thus specializing it is more useful.
> 
> For Valhalla, we are at year 10 and value classes are not there yet, specialization of generics is in my opinion 10 years ahead.
> I'm not complaining, if we had rush value classes, everybody will have regret it, but those things take a lot of time.
> 

Also we may benefit from stack flattening with the boxed primitives. I would hold off on any such explicit specialization (the pressure on Streams was too great, less so for Gatherers IMO). Gatherers may provide a useful way to explore the effectiveness of using primitive boxes that don’t have identity.

Paul.