I think we have everything we need to proceed with a 1.3.0 RC1. I can hopefully prepare it later today and call a vote on it if it all goes well.
On 2025/11/13 05:00:04 kerr wrote: > If anyone is very willing to support Java 8, they could consider > maintaining a 1.4.0 version afterward. Of course, I personally am unwilling > to do so because our internal systems have all been migrated to Java 11 or > Java 21. > > 何品 > > > kerr <[email protected]> 于2025年11月13日周四 12:58写道： > > > I think we should release version 1.3.0 as soon as possible; it's already > > enough. If anyone is willing, they can check what other features we can > > migrate. Otherwise, it's good enough, and we should quickly switch to > > version 2.0.0 and Java 17. This branch is more worthy of our time as > > volunteers. > > > > 何品 > > > > > > kerr <[email protected]> 于2025年11月13日周四 12:48写道： > > > >> I believe we should have all the features of the Akka core. Since this > >> version is already part of Apache 2.0, we should have all the features, > >> allowing us to offer what others have. As for user feedback, I think users > >> only report problems when they encounter them; not everyone visits GitHub. > >> Most people probably just chat with colleagues or complain on messaging > >> apps. > >> > >> Of course, openly speaking, Akka can also pick code from Pekko, > >> complementing each other. Pekko proactively identified several issues in > >> new Scala versions, driving Scala fixes and paving the way for a smooth > >> upgrade of Akka to Scala later. Currently, Akka seems to be focusing > >> primarily on its agentic system, and I've seen an MCP implementation on > >> Pekko. Therefore, to better serve users, we should carefully review new > >> Akka releases. After all, every implementation has a reason. However, since > >> we are not providing a commercial service, we can potentially move faster > >> in terms of binary compatibility; for example, we removed a lot of code in > >> 2.0.0. > >> > >> My personal suggestion is that we align with Akka 2.7.0 in version 1.3.0, > >> avoiding any disruptive changes. Version 2.0.0 should include all the > >> features of Akka 2.7.0, and then prepare a new version, such as 2.1.0, with > >> the features of Akka 2.8.0. Alternatively, we could postpone Pekko 2.0.0 to > >> integrate Akka 2.8.0 features, which would give us more development time. > >> Only in this way can we better serve our users and expand the community. > >> > >> 何品 > >> > >> > >> kerr <[email protected]> 于2025年11月13日周四 12:27写道： > >> > >>> I think 2479 will not make it because it currently breaks Mima. But > >>> https://github.com/apache/pekko/pull/2486 seems ok. > >>> > >>> I think we should get 1.3.0 out soon, because users may switch to Akka > >>> 2.7.0 because of the new behavior api for Java 21's pattern matching. > >>> > >>> https://github.com/CajunSystems/cajun, a new actor library, advertises > >>> that too. > >>> 何品 > >>> > >>> > >>> PJ Fanning <[email protected]> 于2025年11月12日周三 17:37写道： > >>> > >>>> I think PR 2479 is a workable solution. If there are no strong > >>>> objections to it, we could get it into 1.3.0. > >>>> > >>>> Let's delay the RC1 for 1.3.0 for a week or two to allow a discussion > >>>> to happen. > >>>> > >>>> On Wed, 12 Nov 2025 at 09:46, Matthew de Detrich <[email protected]> > >>>> wrote: > >>>> > > >>>> > On another note and on the topic of making a javadsl/scaladsl for > >>>> > ActorSystem, in > >>>> > exploring the option of creating a non-source breaking smooth > >>>> transition > >>>> > from 1.3.0 > >>>> > to 2.0.0 I made a PR at https://github.com/apache/pekko/pull/2479, > >>>> see > >>>> > https://github.com/apache/pekko/issues/2093#issuecomment-3520511719 > >>>> > > >>>> > I think it would be good to make an "executive" decision on the course > >>>> > forward, if > >>>> > we care about ActorSystem in Pekko 1.3.0 being source compatible with > >>>> Pekko > >>>> > 2.0.0 then we would need this PR to be merged for 1.3.0. > >>>> > > >>>> > On the other hand, if we wan't a cleaner API and don't worry about > >>>> source > >>>> > breakage > >>>> > in 2.0.0, that PR isn't needed at all and it would also give us some > >>>> > breathing room as > >>>> > we only need to target Pekko 2.0.0. > >>>> > > >>>> > Thoughts? > >>>> > > >>>> > On Tue, Nov 11, 2025 at 3:31 PM Matthew de Detrich < > >>>> [email protected]> > >>>> > wrote: > >>>> > > >>>> > > > I don't personally believe that we need to port everything from > >>>> Akka > >>>> > > releases as they become available under the Apache license. I would > >>>> > > prefer to concentrate on bug fixes and test coverage. Enhancements > >>>> if > >>>> > > people want them but I don't think we should grab them without > >>>> > > evidence they are wanted by Pekko users. > >>>> > > > >>>> > > I actually disagree here, although within reason. For me, if there > >>>> are > >>>> > > changes > >>>> > > that are isolated and easy to port then we should do that (if > >>>> someone is > >>>> > > willing > >>>> > > to do it), it's better for end users and there is no argument > >>>> against it > >>>> > > aside from > >>>> > > rushing a 1.3.x release. > >>>> > > > >>>> > > I understand that we need more tests and bug fixes but it's not a > >>>> zero sum > >>>> > > game, and > >>>> > > in any case the Akka devs are very diligent in adding tests to any > >>>> > > features that they > >>>> > > implement so porting back changes is not going to change our status > >>>> quo > >>>> > > very much. > >>>> > > > >>>> > > And to close off, I wouldn't rely that much on user feedback when > >>>> it comes > >>>> > > to Pekko > >>>> > > because it's historically not a very good way to gauge what > >>>> features users > >>>> > > want. > >>>> > > Generally speaking people complain when there is a bug/something is > >>>> not > >>>> > > working > >>>> > > (my personal theory for behind this is that its a holdover from how > >>>> Akka > >>>> > > was managed, i.e. > >>>> > > Akka being BDFL and driving the project and our users haven't > >>>> transitioned > >>>> > > to a > >>>> > > community mindset fully). > >>>> > > > >>>> > > Kind regards, > >>>> > > Matthew > >>>> > > > >>>> > > On Sat, Nov 8, 2025 at 8:24 PM PJ Fanning <[email protected]> > >>>> wrote: > >>>> > > > >>>> > >> I don't personally believe that we need to port everything from > >>>> Akka > >>>> > >> releases as they become available under the Apache license. I would > >>>> > >> prefer to concentrate on bug fixes and test coverage. Enhancements > >>>> if > >>>> > >> people want them but I don't think we should grab them without > >>>> > >> evidence they are wanted by Pekko users. > >>>> > >> > >>>> > >> Once Pekko 2.0.0 is out, I don't think we should continue to take > >>>> Akka > >>>> > >> changes over to 1.x unless they fix critical bugs - that they > >>>> should > >>>> > >> only go into 2.x in normal circumstances. > >>>> > >> > >>>> > >> There is a reasonable chance that Akka 2.8.0 changes will become > >>>> > >> Apache licensed before we get to release Pekko 2.0.0. But maybe, it > >>>> > >> might focus our minds to get 2.0.0 complete before then. We could > >>>> then > >>>> > >> just add Akka 2.8.0 stuff in a Pekko 2.x release. > >>>> > >> > >>>> > >> On Sat, 8 Nov 2025 at 20:08, kerr <[email protected]> wrote: > >>>> > >> > > >>>> > >> > I see, but if we want to support 1.4.0, then we will have much > >>>> to port, > >>>> > >> eg, > >>>> > >> > Akka 2.8.0 needs to be ported to Pekko 1.4.x too . > >>>> > >> > And we don't have the same setup, eg, sortImports, Scala > >>>> versions, Java > >>>> > >> > formatter, and Scala formatter, etc., which causes cherry > >>>> picking a huge > >>>> > >> > burden. > >>>> > >> > While porting recently, I had to do many manual sortings to make > >>>> the > >>>> > >> code > >>>> > >> > work with 1.3.x > >>>> > >> > > >>>> > >> > 何品 > >>>> > >> > > >>>> > >> > > >>>> > >> > PJ Fanning <[email protected]> 于2025年11月9日周日 01:43写道： > >>>> > >> > > >>>> > >> > > 1.x releases will support Java 8. > >>>> > >> > > I'm not going to guess what sort of 1.x releases we will need > >>>> but we > >>>> > >> > > will continue to do 1.x releases including some small > >>>> enhancements > >>>> > >> > > until 2.0.0 full release happens. After 2.0.0 is out, I think > >>>> it is > >>>> > >> > > fairly likely that we will only fix bugs in 1.x and this will > >>>> likely > >>>> > >> > > mean only occasional patch releases. > >>>> > >> > > We could easily end up with 1.3.1 or 1.4.0 releases and > >>>> possibly > >>>> > >> beyond. > >>>> > >> > > > >>>> > >> > > On Sat, 8 Nov 2025 at 14:37, kerr <[email protected]> wrote: > >>>> > >> > > > > >>>> > >> > > > Is Pekko 1.3.0 the last release that we plan to support Java > >>>> 8? > >>>> > >> > > > >>>> > >> > > > >>>> --------------------------------------------------------------------- > >>>> > >> > > To unsubscribe, e-mail: [email protected] > >>>> > >> > > For additional commands, e-mail: [email protected] > >>>> > >> > > > >>>> > >> > > > >>>> > >> > >>>> > >> > >>>> --------------------------------------------------------------------- > >>>> > >> To unsubscribe, e-mail: [email protected] > >>>> > >> For additional commands, e-mail: [email protected] > >>>> > >> > >>>> > >> > >>>> > >>>> --------------------------------------------------------------------- > >>>> To unsubscribe, e-mail: [email protected] > >>>> For additional commands, e-mail: [email protected] > >>>> > >>>> > --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
