[External] : Re: New informational JEP: 14: The Tip & Tail Model of Library Development

[Ron Pressler]

> On 5 Nov 2024, at 00:49, Alan Snyder <javalists at cbfiddle.com> wrote:
> 
> I understand the informational part. It is the enhancement part that is questionable.


There was, indeed, an enhancement to the JDK’s development process when T&T was adopted *in the JDK*, but informational JEPs are informational documents that use the name JEP — don’t get stuck on the original meaning of the acronym. It is only feature JEPs — by far the most common JEP type — that actually describe enhancement to the JDK software.

Having said that, developers who ignore the advice in information JEPs — which do accompany fundamental changes in the philosophy behind the JDK’s evolution — will likely have a less-than-ideal experience. In this case, libraries that follow one-size-fits-all will find that they spend more effort for a less valuable result than those that adopt tip & tail, at least partly because they create unnecessary and costly friction between their release model and the JDK’s. We both expect and empathise with scepticism about tip & tail; after all, we had expressed the same scepticism before we experienced the reduction in costs combined with rise in user satisfaction when we actually implemented this model. 

This is natural and will take a while until most libraries adopt this better and easier model (similar scepticism about automated unit tests was widespread among developers, and it took many years for the technique to become common industry practice). Those that do will then wonder — as we did — how they’ve ever put up with other, more arduous and less beneficial release models, such as one-size-fits-all.

— Ron