I support merging now/soon for pekko 1.0.x. It introduces some extra maintenance burden especially to support 2.12 and 3 at the same time. On the other hand, the most risky parts that the branch introduces for Scala 2.x support have already been merged with the move to the latest upstream parboiled2 version. Maintaining the branch longer will only become more difficult. I will hopefully have another look at the state of the branch next week, but I think it should be in good shape. We should do a real merge commit here and not do any rebasing or squashing in this case to preserve the history and attribution as it is.
Johannes On Sat, Apr 15, 2023 at 11:17 AM Matthew Benedict de Detrich <matthew.dedetr...@aiven.io.invalid> wrote: > > So regarding this proposal specifically of merging the scala3 branch into > pekko-http main I am all for it. The risk is low and if there are any > potential issues its better we find them out now rather than later > considering its for a 1.0.x release. There are also performance fixes which > we should reintroduce which have been removed when we moved to upstream > Parboiled2 > > > Granted you likely don't care all that much about what one consumer > thinks, > but i wouldn't be surprised if others are in similar situations. > > > We're in a similar situation to Dave here. Do you have an indication for > how long is left on the scala3 pekko-http support? > > I wouldn't be so rash, we actually care a lot about the users (or at least > I do). The biggest problem we are experiencing is there is a lot of factors > at play which are causing tensions. For example the original plan was to > make Pekko 1.0.0 as close as possible to Akka 2.6 BSL with no behavioural > changes but when then realized there were changes which would be for the > better of the community. One example of such change is updating Jackson > (due to CVE's) which also forced us to upgrade from Scala 3.2. > > Then there are other factors at play such as Scala 3.3, there are extremely > strong arguments by many people (including Scala center/EPFL) that Pekko > should target Scala 3.3 since its a LTS. The core of the bind here really > is binary compatibility/stability. If we release Pekko 1.0.0 with a Scala 3 > version, we are likely going to be stuck with that version for a LOOONG > time considering that we made an agreement that only CVE's/critical fixes > will be backported to 1.0.x branch. My personal overview of the situation > is that at least technically speaking (i.e. aside from the > license/header/legal issues) there isn't much to do. There is a project > with a brief overview of what needs to be done at > https://github.com/orgs/apache/projects/220/views/1 so I would say as a > very broad estimation there is probably 1-2 months of work which should > also line up well with a Scala 3.3 LTS release (that is expected this > month). There is also https://github.com/apache/incubator-pekko/pull/281 > which we should make a discussion on, I will create a discussion thread for > this. > > On a tangential note, a discussion should probably be made about the > evolution of the Pekko project in general wrt binary > compatibility/stability. At the heart of these problems is the expectation > of extreme binary compatibility that is inherited from Akka and I think > there is merit in exploring whether such expectations is the healthiest for > the project in general (i.e. should they be loosened a bit?). > > On Sat, Apr 15, 2023 at 1:33 AM Greg Methvin <gr...@apache.org> wrote: > > > I support this proposal. Scala 3 support is something most people want in a > > Scala library these days, so having it would make the 1.0.0 release feel > > more complete, especially for new users. It would also allow library > > authors to publish new releases using the Scala 3 artifacts as soon as > > possible. > > > > The only real concern is how much it would delay the release. If it did > > cause a delay, I imagine we could put out a milestone release with > > everything except the Scala 3 support, to give people a chance to start > > migrating earlier? > > > > > > > > On Fri, Apr 14, 2023 at 8:37 AM PJ Fanning <fannin...@gmail.com> wrote: > > > > > incubator-pekko release is not dependent on anything in > > > incubator-pekko-http. The original discussion has nothing to do with core > > > pekko. incubator-pekko will be released when they are ready. > > > incubator-pekko-http will be released separately, some time later when it > > > is ready. If you want to discuss incubator-pekko, please start a new mail > > > thread. > > > > > > On Fri 14 Apr 2023, 17:27 Sam Byng, <samb...@microsoft.com.invalid> > > wrote: > > > > > > > We're in a similar situation to Dave here. Do you have an indication > > for > > > > how long is left on the scala3 pekko-http support? > > > > > > > > The positives on our side are that we don't have to wait for pekko > > 1.1.0 > > > > to get pekko-http, and it makes further releases of connectors etc > > > simpler. > > > > However, negatives would be the possible extension of 1.0.0 date. So > > far, > > > > looking at the MR it seems that adding pekko-http scala3 support is not > > > far > > > > off so wouldn't extend the 1.0.0 release too dramatically. > > > > > > > > -Sam > > > > > > > > > -----Original Message----- > > > > > From: Dave Brosius <mebigfat...@gmail.com> > > > > > Sent: Friday, April 14, 2023 2:49 PM > > > > > To: dev@pekko.apache.org > > > > > Subject: Re: [DISCUSS] adding pekko-http scala3 support now for > > v1.0.0 > > > > release > > > > > > > > > > As a future simple consumer of Apache Pekko, i'd would love anything > > > > that gets a published release sooner than later as our corporate > > > governance > > > > is on our necks about using akka (even the > last o/s variant) because > > of > > > > the license change. We have 1 year from the announcement (sept 23) to > > > > resolve. > > > > > > > > > > Granted you likely don't care all that much about what one consumer > > > > thinks, but i wouldn't be surprised if others are in similar > > situations. > > > > > > > > > > On Wed, Apr 12, 2023 at 5:54 AM Nicolas Vollmar <nvoll...@gmail.com> > > > > wrote: > > > > > > > > > > I assume overall there weren't any (major) changes to public APIs for > > > > > Scala 3, so merging it for 1.0.0 would be a small risk, but also > > > > > reduce burden of maintaining the branch and allow to ship Scala 3 > > > > > support across the board with 1.0.0. I'd +1 that. > > > > > > > > > > On Tue, 11 Apr 2023 at 13:26, PJ Fanning <fannin...@apache.org> > > wrote: > > > > > > > > > > > Hi, > > > > > > > > > > > > I'd like to pitch the idea of just merging the pekko-http scala3 > > > > > > support to main branch when it is ready and including this in the > > > > v1.0.0 release. > > > > > > We have already made small-ish changes like using Parboiled jar and > > > > > > upgrading Jackson. > > > > > > > > > > > > The scala3 changes don't make significant changes to the APIs and > > it > > > > > feels > > > > > > like adding the scala3 support now would not make migration from > > > > > > Akka > > > > > HTTP > > > > > > much harder. Akka HTTP has released scala3 support (BSL licensed) > > > > > > but the release seems to have gone smoothly - without much user > > > > complaint. > > > > > Nothing > > > > > > significant had to be documented about the migration to Akka HTTP > > > > > > 10.4 > > > > > [1]. > > > > > > > > > > > > My main reason for supporting an early merge of this is that it > > will > > > > > > save us a whole circle of releases downstream. A scala3 support > > > > > > pekko-http > > > > > > v1.1.0 would lead to new releases for pekko-connectors and other > > > > > downstream > > > > > > projects. > > > > > > > > > > > > I get that we want to make migration to v1.0.0 easy but I don't > > > > > > think the > > > > > > scala3 changes make this significantly harder. > > > > > > > > > > > > If we had made faster progress with the v1.0.0 release then being > > > > > > conservative probably makes sense but now that we still don't have > > a > > > > > > release scheduled, it feels like we might be better off planning to > > > > > > get a slightly bigger v1.0.0 release done and saving ourselves the > > > > > > hassle of having to do a v1.1.0 release for the scala3 changes. > > > > > > > > > > > > [1] > > > > > > > > > > > > > https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdoc. > > > > > akka.io > > %2Fdocs%2Fakka-http%2Fcurrent%2Fmigration-guide%2Fmigration-gui > > > > > de-10.4.x.html%23general-notes&data=05%7C01%7Csambyng% > > 40microsoft.com% > > > > > > > 7C84d3391cb35f40598c2908db3ceefd85%7C72f988bf86f141af91ab2d7cd011db47% > > > > > > > 7C1%7C0%7C638170769415330544%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwM > > > > > > > DAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdat > > > > > a=9kKrhjPaZVGmWaV%2FPcF%2BygzMZjd%2BzXwNpCIuQxyD%2FcY%3D&reserved=0 > > > > > > > > > > > > > > > > > > > > > > > > > > -------------------------------------------------------------------- > > > > > > - To unsubscribe, e-mail: dev-unsubscr...@pekko.apache.org For > > > > > > additional commands, e-mail: dev-h...@pekko.apache.org > > > > > > > > > > > > > > > > > > > > > > > > > --------------------------------------------------------------------- > > > > To unsubscribe, e-mail: dev-unsubscr...@pekko.apache.org > > > > For additional commands, e-mail: dev-h...@pekko.apache.org > > > > > > > > > > > > -- > > Matthew de Detrich > > *Aiven Deutschland GmbH* > > Immanuelkirchstraße 26, 10405 Berlin > > Amtsgericht Charlottenburg, HRB 209739 B > > Geschäftsführer: Oskari Saarenmaa & Hannu Valtonen > > *m:* +491603708037 > > *w:* aiven.io *e:* matthew.dedetr...@aiven.io --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@pekko.apache.org For additional commands, e-mail: dev-h...@pekko.apache.org
