Is it time to move on? Should feign move to java 8?

[velo]

Right now, feign is compatibility is bound to java 6.

Is it time to release feign 10.0 with java 8? (10.x was @spencergibb suggestion)

@kdavisk6 suggested we start a discussion about it.

[codefromthecrypt]

Up to you, but if this changes, please change the group ID and the package ID as it won't be bytecode compatible anymore so definitely a major release. Decisions like this are open, but because the collateral damage extends well beyond those watching issues, just be very cautious if implementing this. We don't want to create work for "innocent bystanders" and changing package and group Id helps with this.

…

On Sat, Mar 17, 2018 at 11:55 AM, Marvin Froeder ***@***.***> wrote: Right now, feign is compatibility is bound to java 6. Is it time to release feign 10.0 with java 8? (10.x was @spencergibb <https://github.com/spencergibb> suggestion) @kdavisk6 <https://github.com/kdavisk6> suggested we start a discussion about it. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#656>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAD613jKbibWPA6Qui4V7k8sL-plo9o7ks5tfIksgaJpZM4SuqAq> .

[codefromthecrypt]

ps @spencergibb <https://github.com/spencergibb> might disagree with me about if changing package names helps or not. It does invalidate other layered tools.. so don't take my word as gospel

[spencergibb]

Spring doesn't change group id or artifact id for minimum Java versions

[codefromthecrypt]

Spring doesn't change group id or artifact id for minimum Java versions

I guess what I mean to ask isn't so much about spring code, but spring cloud impact. When feign is rewritten to java 8 it will not be compatible anymore with java 7. Is this an issue for spring cloud? If so, maybe group id, package. If not, then maybe unnecessary. At any rate the android support statements should be dropped if this happens, so yeah just update that part. In other projects I've avoided bumping to java 8 (and dealing with this) via retrolambda. Anyway there are a few folks who will be invalidated. For example, anyone using feign inside a metrics library as an exporter, it will invalidate the ability to use it (which might be fine as they just put a <10 update comment in their version matrix).

[velo]

the android support statements should be dropped if this happens.
In my mind, android had java 8 support
https://developer.android.com/studio/write/java8-support.html
Am I missing something?

Yeah, https://github.com/orfjackal/retrolambda seems a good option.
We could even build the official packages with java 8 and offer retrofitted java 6 jars using classifier.

[codefromthecrypt]

In my mind, android had java 8 support https://developer.android.com/studio/write/java8-support.html Am I missing something? There's an SDK version level concern. For example, the current code is

compatible with older versions https://github.com/Netflix/denominator/blob/master/example-android/build.gradle#L14

[spencergibb]

Nope, spring cloud (and framework and boot) are now all baselined to Java 8.

[kdavisk6]

By moving to Java 8, what will we gain? (Not sarcastic, honest question).

[spencergibb]

Use of java 8 language and APIs inside core feign.

[velo]

By moving to Java 8, what will we gain?
Can squash feign-java8 into feign-core, which is meh level of awesomeness

Allow us to merge this #638
Right now, the PR introduces a java 8 change to feign-hystrix

[kdavisk6]

I understand that #638 wants to add CompleteableFuture, which sounds really good from a user perspective. Are there specific areas within feign, internally now that would benefit immediately from Java 8 API's? Are we considering this a spring board for a complete refactor or simply to support things like #638? I'm trying to understand how far we would like to take it.

[kdavisk6]

Just to be clear, I'm all for moving to Java 8. I am trying to get an idea on how much do we need to change versus should change to get there.

[kdavisk6]

Is this something we still want to pursue? If so, what do we need to get started?

[velo]

I think we should consider feign 10 with java 8.

I don't have any features in mind, but that will open doors for sure.
May be, for feign 10 we publish java 8 and use retrofeign to keep publishing java 6 libraries, with plans to remove java 6 support on feign 11.

How does that sound like?

[masc3d]

but that will open doors for sure.

it will certainly close door to being compatible with android < sdk24 very quickly @velo

[velo]

@masc3d as you can see, this was done 3 months ago.
Do you think retrofeign would help?

[masc3d]

Do you think retrofeign would help?

@velo I assumed this was rather about retrofitting async support #644

[velo]

Meant to type retrolambda