For example , if I ran sbt package and sbt will generate a jar name like project_2.11-version.jar, how can I modify this name to a random name?
Here is what the docs say:
The generated artifact name is determined by the artifactName setting.
This setting is of type (ScalaVersion, ModuleID, Artifact) => String.
And the default implementation is:
artifactName := { (sv: ScalaVersion, module: ModuleID, artifact: Artifact) =>
artifact.name + "-" + module.revision + "." + artifact.extension
}
You can copy this code into your build.sbt or Build.scala file and change how it constructs the artefact name. For example:
artifactName := { (sv: ScalaVersion, module: ModuleID, artifact: Artifact) =>
java.util.UUID.randomUUID.toString + "." + artifact.extension
}
Related
I want to create a custom sbt task that use publishLocal in Docker, login in ecr, and publishLocal in Docker.
Also note that I modified "publishLocal in Docker" settings only on custom task.
I tried writing the code bellow but it did not work.
val releaseDev = taskKey[Unit]("push docker image")
releaseDev := (push in ecr).value
releaseDev := releaseDev.dependsOn(publishLocal in Docker).value
releaseDev := (projectSettings ++ Seq(repositoryName in ecr := (packageName in Docker).value + "-stg" + ":" + (version in Docker).value))
Here is how to trigger existing tasks from your custom one. From your question, you want them to be triggered in some particular sequence, so here is what you need to do
lazy val releaseDev = taskKey[Unit]("push docker image")
releaseDev := Def.sequential(push in ecr, publishLocal in Docker).value
Now when you trigger releaseDev it will first turn push in ecr then publishLocal in Docker.
Note: don't forget to add the lazy in your taskKey definition since it may lead to some strange initialization issues.
Thank you marios.
I tried writing the code below and task dependencies were properly handled.
However repositoryName is used that one of project settings.
// docker publish settings
import com.amazonaws.regions.{Region, Regions}
region in ecr := Region.getRegion(Regions.AP_NORTHEAST_1)
repositoryName in ecr := (packageName in Docker).value + ":" + (version in Docker).value
localDockerImage in ecr := (packageName in Docker).value + ":" + (version in Docker).value
// Publisher Setting
//~~~~~~~~~~~~~~~~~~~
import ReleaseTransformations._
releaseProcess := Seq[ReleaseStep](
checkSnapshotDependencies,
inquireVersions,
runClean,
runTest,
setReleaseVersion,
commitReleaseVersion,
tagRelease,
ReleaseStep(state => Project.extract(state).runTask(publishLocal in Docker, state)._1),
ReleaseStep(state => Project.extract(state).runTask(login in ecr, state)._1),
ReleaseStep(state => Project.extract(state).runTask(push in ecr, state)._1),
setNextVersion,
commitNextVersion,
pushChanges
)
lazy val releaseDev = taskKey[Unit]("push docker image")
releaseDev := Def.sequential( publishLocal in Docker, login in ecr, push in ecr).value
(repositoryName in ecr) in releaseDev := (packageName in Docker).value + "-stg" + ":" + (version in Docker).value
So I'm trying to use sbt-release, and I'm having issues where it's publishing the artifact to my snapshot repository, and not the release repository.
val Organization = "com.mycompany"
val Name = "My Service"
val Version = "0.1-SNAPSHOT"
...
settings = Defaults.coreDefaultSettings ++ Seq(
name := Name,
organization := Organization,
version := Version,
scalaVersion := ScalaVersion
...
assemblyJarName in assembly := s"my-service-${Version}.jar",
...
)
publishTo := {
val nexus = "my.nexus.url.com/repositories/"
if (isSnapshot.value)
Some("snapshots" at nexus + "snapshots/")
else
Some("releases" at nexus + "releases/")
},
credentials += Credentials(Path.userHome / ".ivy2" / ".credentials")
If I remove the -SNAPSHOT from the version, then it publishes it to the correct repository, but shouldn't sbt-release be telling it to do that by itself?
Also if I get rid of the if (isSnapshot.value) then sbt publish will also publish to the wrong repository.
If I could get some help on this, I would really appreciate it.
It was the version I had here. It was over riding the version.sbt which is where 0.1-SNAPSHOT should be stored.
I am in the middle of upgrading Nexus version. As part of the process I've set up a new Nexus instance which will run in parallel with the older Nexus instance.
While migrating to the new instance I want to thoroughly test and vet the new instance before pulling the plug on older instance. This requires me to temporarily modify the publish workflow in such a way that sbt publishes the artifacts to both the Nexus instances.
I highly doubt the following code will actually work:
publishTo <<= (version) {
version: String =>
if (version.trim.endsWith("SNAPSHOT")) Some("snapshots" at "http://maven1.dev.net:8081/nexus/content/" + "repositories/snapshots/")
else Some("releases" at "http://maven1.dev.net:8081/nexus/content/" + "repositories/releases/")
},
credentials += Credentials("Sonatype Nexus Repository Manager", "maven1.dev.net", "release-eng", "release"),
publishTo <<= (version) {
version: String =>
if (version.trim.endsWith("SNAPSHOT")) Some("snapshots" at "http://maven2.dev.net:8081/nexus/content/" + "repositories/snapshots/")
else Some("releases" at "http://maven2.dev.net:8081/nexus/content/" + "repositories/releases/")
},
credentials += Credentials("Sonatype Nexus Repository Manager", "maven2.dev.net", "release-eng", "release"),
I also tried looking into a plugin called sbt-multi-publish but I couldn't compile and use it, either.
With Commands and How to change a version setting inside a single sbt command? I could define a new command - myPublishTo - that changes publishTo setting before executing the original publish task:
def myPublishTo = Command.command("myPublishTo") { state =>
val extracted = Project.extract(state)
Project.runTask(
publish in Compile,
extracted.append(List(publishTo := Some(Resolver.file("file", target.value / "xxx"))), state),
true
)
Project.runTask(
publish in Compile,
extracted.append(List(publishTo := Some(Resolver.file("file", target.value / "yyy"))), state),
true
)
state
}
commands += myPublishTo
With this, execute myPublishTo as any other command/task.
You could also define a couple of aliases - pxxx, pyyy and pxy - in build.sbt that would execute a series of commands using ;.
addCommandAlias("pxxx", "; set publishTo := Some(Resolver.file(\"file\", target.value / \"xxx\")) ; publish") ++
addCommandAlias("pyyy", "; set publishTo := Some(Resolver.file(\"file\", target.value / \"yyy\")) ; publish") ++
addCommandAlias("pxy", "; pxxx ; pyyy")
In sbt console you can execute them as any other commands/tasks.
[sbt-0-13-1]> alias
pxxx = ; set publishTo := Some(Resolver.file("file", target.value / "xxx")) ; publish
pyyy = ; set publishTo := Some(Resolver.file("file", target.value / "yyy")) ; publish
pxy = ; pxxx ; pyyy
[sbt-0-13-1]> pxy
[info] Defining *:publishTo
[info] The new value will be used by *:otherResolvers, *:publishConfiguration
[info] Reapplying settings...
[info] Set current project to sbt-0-13-1 (in build file:/Users/jacek/sandbox/so/sbt-0.13.1/)
...
[info] published sbt-0-13-1_2.10 to /Users/jacek/sandbox/so/sbt-0.13.1/target/xxx/default/sbt-0-13-1_2.10/0.1-SNAPSHOT/sbt-0-13-1_2.10-0.1-SNAPSHOT-javadoc.jar
[success] Total time: 1 s, completed Jan 9, 2014 11:20:48 PM
[info] Defining *:publishTo
[info] The new value will be used by *:otherResolvers, *:publishConfiguration
[info] Reapplying settings...
...
[info] published sbt-0-13-1_2.10 to /Users/jacek/sandbox/so/sbt-0.13.1/target/yyy/default/sbt-0-13-1_2.10/0.1-SNAPSHOT/sbt-0-13-1_2.10-0.1-SNAPSHOT-javadoc.jar
[success] Total time: 0 s, completed Jan 9, 2014 11:20:49 PM
This is an old question, but the problem persists. I tried to revive sbt-multi-publish, but it's really old (sbt-0.12) and uses some sbt internals that are hard to deal with. So I took another approach and wrote a new plugin: sbt-publish-more.
It doesn't involve any on-the-fly settings changing or custom commands like the other answer.
After you add the plugin, just set resolvers you want to publish to (taking your code as an example):
publishResolvers := {
val suffix = if (isSnapshot.value) "shapshots" else "releases"
Seq(
s"Maven1 ${suffix}" at s"http://maven1.dev.net:8081/nexus/content/repositories/${suffix}/",
s"Maven2 ${suffix}" at s"http://maven2.dev.net:8081/nexus/content/repositories/${suffix}/"
)
}
And call publishAll task, it will publish to both repositories.
You can also publish to different repositories with different configurations. Check usage docs for details.
I am very new to SBT and need to create a RPM package for one of my project. The RPM contains only 1 file, which is a one-jar created by the sbt-onejar plugin). I want to use sbt-native-packager plugin and have created a Packagin.scala file under the /project directory like this:
object Packaging {
val settings: Seq[Setting[_]] = packagerSettings ++ deploymentSettings ++ mapGenericFilesToLinux ++ Seq(
maintainer := "Team",
packageSummary := "Summary",
packageDescription := """Description""",
mappings in Universal += {
file("target/scala-2.10/projectname_2.10-0.1-one-jar.jar") -> "/opt/projectname/projectname-0.1.jar"
},
linuxPackageMappings in Rpm <+= (baseDirectory) map { _:File =>
(packageMapping(file("target/scala-2.10/projectname_2.10-0.1-one-jar.jar") -> "/opt/projectname/projectname-0.1.jar")
withUser "someusr" withGroup "somegroup" withPerms "0755")
},
name in Rpm := "projectname",
version in Rpm <<= version apply { sv => sv split "[^\\d]" filterNot (_.isEmpty) mkString "." },
rpmRelease := "1",
rpmVendor := "Vendor",
rpmUrl := Some("url"),
rpmGroup := Some("group"),
rpmLicense := Some("BSD")
)
}
1) I don't want to hardcode the file names. Instead of having "target/scala-2.10/projectname_2.10-0.1-one-jar.jar" I need a way to use existing SettingKey's, i.e. target + "scala-" + scalaVersion + "/" + name + "_" + scalaVersion + "-" + version + "-one-jar.jar" - how do you do this=
2) For the value rpmRelease := "1" I want to use a System property, i.e. in Maven I would do ${rpm.buildNumber}, how does that work in SBT?
3) Is there anything I should do better in regards to the sbt-native-packager plugin?
1) You should always use task output in sbt rather than raw filesystem lookups. Because sbt has parallel execution, if you don't put an explicit dependency on the output of a task, then you have no guarantee that a file will be created before you run your task.
In that vein you want to change your package mappings line to be something like this:
mappings in Universal += {
oneJar.value -> "/opt/projectname/projectname-0.1.jar"
},
Where the oneJar key is defined in the onejar plugin.
2) Sbt just uses scala for the build language, so you can grab system properties similarly (but please also provide a default):
rpmRelease := Option(sys.props("rpm.buildNumber")) getOrElse "1"
3) Right now you're defining a generic package and redefining the same file in the Rpm with a different user. The mapGenericFilesToLinux settings still lack a few customizations, but if you're not creating universal distributions, you should be able to drop that bit of settings and instead directly configure your linux package:
linuxPackageMappings in Rpm <+= (oneJar) map { jar:File =>
(packageMapping(jar -> "/opt/projectname/projectname-0.1.jar")
withUser "someusr" withGroup "somegroup" withPerms "0755")
},
Sbt always does full rebuild because of modified binary dependency: rt.jar:
[debug]
[debug] Initial source changes:
[debug] removed:Set()
[debug] added: Set()
[debug] modified: Set()
[debug] Removed products: Set()
[debug] Modified external sources: Set()
[debug] Modified binary dependencies: Set(C:\Program Files\Java\jdk1.6.0_37\jre\lib\rt.jar)
[debug] Initial directly invalidated sources: Set()
Obviously, rt.jar wasn't changed, its created/accesses/modified dates are the same and quite old.
It's in c:\Program Files\Java\jdk1.6.0_37\jre\lib\
JAVA_HOME is set as C:\Progra~1\Java\jdk1.6.0_37
JAVA_HOME/bin is in PATH.
Any ideas why sbt thinks rt.jar was changed?
I had JAVA_HOME set to C:\Progra~1\Java\jdk1.6.0_37 and sbt resolved the dependency as C:\Program Files\Java\jdk1.6.0_37\jre\lib\rt.jar. Sbt uses java.io.File.equals() which is not correct in this case:
In sbt/compile/inc/Incremental.scala:
def externalBinaryModified(entry: String => Option[File], analysis: File => Option[Analysis], previous: Stamps, current: ReadStamps)(implicit equivS: Equiv[Stamp]): File => Boolean =
dependsOn =>
analysis(dependsOn).isEmpty &&
orTrue(
for {
name <- previous.className(dependsOn)
_ = println("Name: " + name)
e <- entry(name)
_ = println("entry: " + e)
} yield {
val resolved = Locate.resolve(e, name)
println("resolved: " + resolved)
println("dependsOn: " + dependsOn)
println("resolved != dependsOn: " + (resolved != dependsOn))
(resolved != dependsOn) || !equivS.equiv(previous.binary(dependsOn), current.binary(resolved))
}
)
gives the following output:
Name: java.lang.Object
entry: c:\Progra~1\Java\jdk1.6.0_37\jre\lib\rt.jar
resolved: c:\Progra~1\Java\jdk1.6.0_37\jre\lib\rt.jar
dependsOn: c:\Program Files\Java\jdk1.6.0_37\jre\lib\rt.jar
resolved != dependsOn: true
Thus, sbt always thinks rt.jar was changed.
Workaround would be setting JAVA_HOME to "c:\Program Files\Java\jdk1.6.0_37"
Solution would be to do:
resolved.getCanonicalPath != dependsOn.getCanonicalPath