Scala3 Play3 Apache Pekko migration#2595
Open
mathieuancelin wants to merge 71 commits into
Open
Conversation
- sbt 1.7.2 → 1.12.5 ; sbt-plugin com.typesafe.play 2.8.19 → org.playframework 3.0.10 - plugin bumps (scalafmt/revolver/bloop/pgp/ci-release) + sbt-assembly 2.3.1 (remplace project/assembly.sbt 0.14.6) - build.sbt : Scala 3.8.3, play-json 3.0.6, akka→pekko 1.6.0 / pekko-http 1.3.0 / connectors, jackson 2.21.3, macwire 2.6.7, rediscala fork 2.0.2, scaffeine 5.3.0, sangria 4.2.18, diffson 4.7.0, wasm4s 5.0.3, pulsar4s-pekko-streams 2.10.0, scalatestplus-play 7.0.2, censorinus via for3Use2_13 - on garde toutes les libs Java aux versions actuelles (netty/reactor-netty, k8s, lettuce, vertx-pg, bouncycastle, opentelemetry…) pour éviter des changements de code hors-migration - drop kaleidoscope (réécrite en Phase 3) et swagger-scala-module (pas de build Scala 3) ; on garde swagger-core-jakarta (Java) - suppression du jar akka-stream patché (le fix TLS 1.3 est dans Pekko 1.6 upstream) et du jar scala-schema inutilisé ; hook Twirl adhocExtensions pour Scala 3.8 ; drop PlayAkkaHttp2Support Résolution des dépendances (sbt update) verte.
- app/ + test/: `akka.` → `org.apache.pekko.` (855 imports, ~315 files)
- alpakka → pekko connectors: `akka.stream.alpakka` → `org.apache.pekko.stream.connectors` (s3)
- `akka.kafka` → `org.apache.pekko.kafka`
- app/tcp/tcputils.scala: `package akka` → `package org.apache.pekko` (keeps in-package access
to NotUsed/Done and stream internals, no more patched akka-stream)
- app/utils/akka.scala: reflective field name mangled for the pekko package
(akka$http$… → org$apache$pekko$http$…) for the open-charset json media type hook
- conf: actorsystem blocks `akka {` → `pekko {`, `${akka.version}`/`${akka.http.parsing.*}`
→ `${pekko.*}`, `play.server.akka` → `play.server.pekko`, pekko.version "2.6.20" → "1.6.0",
logback akka logger → org.apache.pekko ; kept otoroshi keys (enforce-akka, OTOROSHI_AKKA_*
env vars, "Otoroshi-akka" user-agent, metadata keys)
Otoroshi's own Akka-named identifiers/strings (useAkkaHttpClient, AkkaWsClientRequest,
"otoroshi-core-use-akka-http-client", logger names…) are intentionally left unchanged.
- kaleidoscope (macro-based Scala 2 lib, unusable from Scala 3): add a tiny non-macro shim otoroshi.utils.KaleidoscopeShim that re-implements the old `r"...$name@(regex)"` pattern interpolator with String captures, and swap `import kaleidoscope._` → the shim in the 7 files that used it. All ~117 `case r"…"` call-sites (incl. el.scala's 106) stay unchanged. - TemplatesController.templateSpec: `scala.reflect.runtime.universe` / `typeOf` don't exist at runtime in Scala 3 → derive event field paths with plain Java reflection (Product + declared fields, unwrap parameterized fields, visited-set against cycles); `typeOf[T]` → `classOf[T]`. - Env.scala: drop the SwaggerScalaModelConverter registration (swagger-scala-module has no Scala 3 build and pulls Akka); api.scala already falls back to a generic schema.
…plicit types - build.sbt: add -source:3.0-migration so Scala 2.13 syntax (e.g. un-parenthesized typed lambda params) is accepted as warnings instead of errors — keeps the diff minimal - backtick `export` / `enum` used as identifiers (now hard keywords in Scala 3) - add explicit types to ~400 `implicit (lazy) val NAME = rhs` (Scala 3 requires them): ExecutionContext / Materializer / Scheduler / ActorSystem / OFormat / Format / Env, fully-qualified to avoid import churn Compile errors: 1874 → 566.
… fixes
- annotate the lambda param of sync `XxxAction { ctx => ... }` and `actionBuilder { req => }`
call-sites: Play 3's ActionBuilder.apply(block) is ambiguous with apply(bodyParser) in Scala 3
because BodyParser is a SAM (RequestHeader => Accumulator) — the param type disambiguates
- `.mapValues(f)` → `.mapValues(f).toMap` (Scala 2.13 returns a lazy MapView, rejected where a Map
is required, e.g. JsObject(...))
- `Env.metrics`: annotate `: Metrics` — a member overriding HasMetrics.metrics without an explicit
type now inherits the parent type (TimerMetrics) in Scala 3, hiding histogramUpdate/counterInc/…
- script.scala class scans: `.map(_.getName)` → `.map(_.getName).toSeq` (asScala yields mutable.Buffer)
- adminEntityValidators parsing: `.collect{…}` → `.collect{…}.toSeq` (JsArray.value is IndexedSeq,
union-typed away from the expected Map[String, Seq[JsonValidator]])
…er, JsonValidator toSeq
- implicit-first modifier order (implicit private val …) now also gets explicit types (netty server)
- brace-form .mapValues { … } → .toMap (Scala 2.13 MapView), mirroring the paren-form fix
- SAMLClient: import scala.jdk.CollectionConverters._ (asScalaBufferConverter is not in scala.jdk)
Compile errors: 347 → 308.
…r paren/parser forms
- scala.collection.mutable.MutableList (removed in 2.13) → ListBuffer
- reactor-netty .secure(spec => …): make the lambda Unit-returning for the Java Consumer SAM
- annotate remaining ambiguous ActionBuilder forms: ApiAction(ctx => …) paren form and
BackOfficeActionAuth(parse.json) { ctx => … } parser form
Compile errors: 308 → 278.
- treerouter NgMatchedRoutes(routes,…): routes is a mutable ListBuffer → .toSeq for immutable Seq - chains ending in .getOrElse(Seq.empty[X]) whose .map branch yields collection.Seq (JsArray.value) no longer conform to the immutable Seq field type → append .toSeq Compile errors: 278 → 233.
…to-tupling, WSResponse.body[String] - import scala.jdk.CollectionConverters._ (named converters like mapAsScalaMapConverter are not in scala.jdk) - response.scala: rename lambda param r → mr to avoid clash with the KaleidoscopeShim interpolator - restore explicit tuples for println(...) (Scala 3 removed auto-tupling) - resp.body → resp.body[String] (WSResponse.body BodyReadable is ambiguous in Scala 3)
…onfig type Compile errors: 233 → (see next).
…ollection.IndexedSeq, ClassTag Array.empty[String]
… error() collection.Seq param
…ixes EnhancedRow implicit Logger)
…g redis: ReactivePgRedis
…nents, PekkoSSLConfig, breakOut→toSet, Right/rbac/polyfill/cookies toSeq
…llect(), more toSeq (oauth/workflow/teams/analytics), script.scala JsError collection.Seq
….collect toSeq (oidc/EntityFiltering/OpenapiToJson)
…wed by otoroshi.utils.Regex), gzip breakOut, filterKeys/coraza toSeq
…param, izanami response type, sangria InterfaceTypeDefinition interfaces=
…iceArrayWithArgs collection.Seq, OpenapiToJson/oidc/sources/netty toSeq, coraza wasm host fns
…s JavaConverters, netty request varargs toSeq
…ing], api.scala headers immutable, graphql validateQuery args, html applyOnIf explicit lambda, sources/entities toSeq
…If existential, graphql untag urlArg, vertx pool.close handler
…scription, kubernetes body[String]/toSeq, coraza immutable Seq
…ubs, coraza Seq import, xml class not case-class, sangria validateQuery 4-arg
… Play 3 / Pekko (1874 → 0 errors)
(cherry picked from commit 501e17d)
(cherry picked from commit 3b2e86c)
(cherry picked from commit 513071d)
(cherry picked from commit 855a0d7)
(cherry picked from commit f5032a2)
The cherry-picked test from release 17.17.0 used the old ScalaTest 3.0
API (org.scalatest.{MustMatchers, WordSpec}) which no longer exists in
the ScalaTest 3.2+ used on this branch. Switch to AnyWordSpec + must.Matchers
to match the migrated tests convention.
Reconcile the branch with master so PR #2595 becomes mergeable. Conflict resolution: - el.scala, oidc.scala, apikey.scala, elasticsearch.js: kept the Scala 3 branch versions (the release fixes were already backported here; the only differences were scalafmt/prettier formatting). - ElasticWritesAnalyticsSpec.scala: kept the Scala 3 version (ScalaTest 3.2 AnyWordSpec/must.Matchers API), master's used the removed ScalaTest 3.0 API. - build.sbt: took master's version 17.18.0-dev. Verified: sbt compile and sbt Test/compile both succeed.
Removed the call to ElasticWritesAnalytics.initialized for compatibility with newer Elastic versions.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Of course AI assisted