Hi all,

There have been six +1s (including myself) and no negatives, so let's do this.

---

For Robert's question, I did some investigation for use of lambdas as config key values. They don't persist correctly unfortunately [1], so we shouldn't use them anywhere that could end up in persisted state.

---

I suggest we separate this into multiple stages (and ensure that discussion of later stages does not block the earlier stages):

 * Stage one: switch to Java 8 as "official target"
    1. Update our jenkins etc to always build/run with Java 8.
    2. All developers/testers switch to Java 8 locally.
    3. Add to the next release notes that Java 7 is no longer supported.
 * Stage two: update our poms so they don't compile/check for Java 7
   compatibility [2,3,4].
 * Stage three: update our docs to say not to use lambdas for config
   keys (similar to what we already say for anonyous inner classes [5])
 * Stage four: developers free to use Java 8 language features *where
   it does not affect APIs or end up in persisted state*
 * Stage five: discuss/improve use of Java 8 (e.g. perhaps fix [1];
   should we keep using guava classes that are now in Java 8's
   java.util.function.* ?)

---

Geoff: if you're volunteering, it would be awesome if you could pick up stage 1.1 (and then stage 2).

Aled

[1] https://issues.apache.org/jira/browse/BROOKLYN-448
[2] https://github.com/apache/brooklyn-server/blob/master/pom.xml#L91
[3] https://github.com/apache/brooklyn-server/blob/master/parent/pom.xml#L557-L558 [4] https://github.com/apache/brooklyn-server/blob/master/parent/pom.xml#L950 [5] https://brooklyn.apache.org/v/latest/ops/persistence/index.html#writing-persistable-code


On 06/03/2017 14:35, Geoff Macartney wrote:
hi all,

any more thoughts on this?   What concrete steps do we want to take, and
when?  What can I do to help out?

Plus, has anyone any views on Robert's question?

cheers
Geoff




On Mon, 27 Feb 2017 at 14:21 Robert Moss <[email protected]> wrote:

Aled,

WRT Java 8 features.  Is it still recommended to use named inner classes
over anonymous inner classes? Can we now simplify those areas to use
lambdas?  There are a number of classes now built into Java that were
previously provided by Guava etc. e.g. Optional, Predicate - should we
start using the Java 8 classes instead where possible?

Robert

On 27 February 2017 at 13:08, Aled Sage <[email protected]> wrote:

Hi all,

I propose we switch to Java 8 for Apache Brooklyn (i.e. require Java 8;
remove support for using Java 7 when running Brooklyn).

This will make our testing simpler, improving coverage (e.g. we don't
currently test everything with Brooklyn running both on Java 7 and Java
8).
It will make our lives easier as developers (e.g. don't need to worry
about
Java 7 versus Java 8 when compiling, running tests and manually testing
Brooklyn; and we can use Java 8 features).

Java 7 reached end of life in April 2015 [1]; Java 8 was released March
2014.

We can do this in three stages:

  * Switch to Java 8 as "official target"
      o Update our jenkins etc to always build/run with Java 8.
      o All developers/testers switch to Java 8 locally.
      o Add to the next release notes that Java 7 is no longer supported.
  * Update our poms so they don't compile/check for Java 7 compatibility
    [3,4,5].
  * Developers free to start using Java 8 language features!

Stage 1 would be the bare minimum for the next release; I think we should
do all three stages before the 0.11.0 release.

Aled

[1] https://java.com/en/download/faq/java_7.xml
[2] https://www.infoq.com/news/2015/05/Oracle-Ends-Java-7Public-Updates
[3] https://github.com/apache/brooklyn-server/blob/master/pom.xml#L91
[4] https://github.com/apache/brooklyn-server/blob/master/parent
/pom.xml#L557-L558
[5] https://github.com/apache/brooklyn-server/blob/master/parent
/pom.xml#L950



Reply via email to