Hacker News Reader

Comment by syngrog66 | Hacker News Reader

I refuse to believe Java has gone through 20 major revisions since 1.0

I remember when 1.5 came out and Sun's marketing folks insisted it be called Java 5. I think thats when they jumped the shark, and haven't corrected course back since, versioning-wise

And I wonder if someone keeps track of what the proper traditional Java major version should be now. I'd guess its on 2.x or 3.x at best

View on HN

If you keep to the standard edition platform and never used internal APIs, you could argue Java never broke backwards compatibility, effectively having a 1. before every version number.

In practice, you could say that Java had only one major version bump: from 8 to 9, when it closed down internal APIs with Jigsaw and gave away Java EE to become Jakarta.

Sure, that's easy. The answer is 21. You don't get to write versioning rules for other projects and there is no golden standard.

Is there some kind of objective standard for version numbers I’m not aware of? Maybe you could call only LTS versions “major versions” if you wanted.

There’s https://semver.org/, which Java unfortunately has not chosen to follow. As joaonmatos pointed out, 8->9 was the most painful major version change, and I think there have been a couple others when deprecated features were (or will be) fully removed.

Projects that follow semver in theory often don't actually follow it so why would we expect everyone else to?

I think the attempt communicates a lot of useful info and succeeds pretty often. I agree they can’t guarantee it, as with any bugs.

It’s fine. If your biggest gripe with something made by a corporation is its name then you will always be unhappy.

After 8.0 then started to release very quickly. I'd say, we're now with Java 10u30 by old standards (or 1.10u30 by even older standards)

>And I wonder if someone keeps track of what the proper traditional Java major version should be now. I'd guess its on 2.x or 3.x at best

What does that even mean?

Before Java 5, Java used 1.x scheme for its major releases (Java 1.0, 1.1, 1.2, 1.3, 1.4). Java 5 was both Java 1.5 and Java 5 and was marketed as Java 5. Since then Java switched to new versioning scheme. Some people apparently are salty about that. I'm salty as well, I'd prefer semver, but Google Chrome 114 probably made it too seductive for marketing to use this crazy scheme, so here we are, with releases like Java 20 which add nothing new.

JDK 20 introduced the Scoped Values JEP as a preview, and also had significant changes to the Panama Foreign Function & Memory (FFM) preview API:

- Scoped Values: https://openjdk.org/jeps/429

- Panama FFM 2nd: https://openjdk.org/jeps/434

The changes in the Panama API were so drastic between 19 and 20 that any code you had written would no longer work. Which is the point of interim releases like these.

I happen to have a lot of experimental Panama code that has been continually evolving with these non-LTS releases since the initial release.

I think that preview features are not something that any sizeable chunk of developers will use. So while technically you're correct, I still consider Java 20 to be pretty minor release.

That's fair, and I'm inclined to agree.

Just wanted to point out there wasn't _really_ nothing new, even if it might have been somewhat pedantic.

Well, they were going with 1.5, 1.8 because the 5 and 8 are minor version numbers as it's all backwards compatible.

There is no universal standard for “major revision” so this is fine.

The only seeming standard is that big integers are bad. ;) But don’t tell Chrome that.

>I refuse to believe Java has gone through 20 major revisions since 1.0

What's your take on Chrome and Firefox?

Reply on news.ycombinator.com