Skip to content

SBT workflow plugin for development, testing and publishing of multiple project repositories simultaneously.

License

Notifications You must be signed in to change notification settings

storm-enroute/mecha

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ScalaDoc API

Build Status

Build Status

Maven Artifact

Mecha -- SBT Plugin That Automates Development Workflow

An SBT plugin that allows you to work on multiple interdependent projects at the same time, as if they were a single project.

Ask yourself:

  • Did you ever notice yourself simultaneously working on 10 different projects that depend on each other?

  • Is publishing a snapshot artifact from one project to compile another project slowing you down and killing your productivity?

  • You feel your projects are bitrotting because they don't live in the same repository?

  • Did you ever want to simultaneously work on different Git code repositories as if they were one big project, and commit changes to all of them at the same time?

  • Are automated nightlies and documentation publishing something you crave for?

  • Are you looking for an easy way to configure the project when it's first checked out?

  • Looking for a faster edit-refresh cycle?

  • Or you want to quickly deploy your project over SSH to a remote server?

If the answer is yes to any of the above, you're at the right place. Mecha is an SBT plugin that aggregates all the SBT projects that you specify into one big project, which you can then compile as if they were one project. You can branch/merge/commit changes in all the repositories simultaneously, pull/push to their respective origins with a single SBT command, or automatically maintain any number of downstream mirrors.

Mecha is an ideal solution when you want to compile your projects against the latest version of fast moving dependencies, as the source codes of all your projects get pulled, compiled, committed, and pushed together. Guaranteed to boost productivity!

So, how does all this work?

Mecha uses the concept of a super-repository. A super-repository is a repository in which there are multiple source controlled repositories for different projects. Repositories with the super-repository can have dependencies on each other, can depend on each others' source code, or Maven artifacts. All the repositories can be worked on, built, tested, and committed simulatenously.

Mecha puts repositories into subdirectories of the super-repository. You can choose which subdirectories to track while you work, or instead depend on them through pre-published Maven artifacts.

What are the features, exactly?

In short:

  • supports super-repositories that contain multiple child repositories
  • bulk pull, branch, merge, push, and mirroring for child repositories
  • optional tracking for child repositories
  • automatic source/library dependency configuration in child repositories
  • automatic configuration file generation
  • automated remote deployments over SSH
  • automated documentation deployments to remote Git repositories (ideal for GH pages!)
  • automated benchmark deployments in case you're using ScalaMeter

In what follows, we describe the main features of Mecha on an example project.

Super-Repo Configuration

Let's take a look at an example super-repository -- storm-enroute/examples. If you look inside storm-enroute/examples, you will see a number of different example projects. The example project we are interested in is called mecha-super-repo. It is used to aggregate other projects.

We took four steps in creating this super-repo.

  1. Add the mecha plugin to your build. Create the project/plugins.sbt file and at the following:

     resolvers ++= Seq(
       "Sonatype OSS Snapshots" at
         "https://oss.sonatype.org/content/repositories/snapshots",
       "Sonatype OSS Releases" at
         "https://oss.sonatype.org/content/repositories/releases"
     )
    
     addSbtPlugin("com.storm-enroute" % "mecha" % "0.2")
    

    This will add the plugin to the super-repo SBT build. If necessary, replace 0.2 with the proper Mecha version.

  2. Import the Mecha package contents:

     import org.stormenroute.mecha._
    

    This will make all Mecha stuff visible.

  3. Define a super-repo build in project/Build.scala as in the following example:

     object MechaSuperRepoBuild extends MechaSuperBuild {
    
       def definition(): Project = superProject
    
       lazy val mechaSuperRepoSettings = defaultMechaSuperSettings ++ Seq(
         name := superName,
         scalaVersion := "2.12.1",
         version := "0.1",
         organization := "com.storm-enroute",
         libraryDependencies ++= Seq()
       )
       
       override val superName = "super-storm-enroute"
       override val superSettings = mechaSuperRepoSettings
     }
    

    The values of particular importance here are superName and superSettings.

  4. Reference build definition from build.sbt as in the following example:

     lazy val `super-storm-enroute` = MechaSuperRepoBuild.definition()
    
  5. Last, create a repos.conf file in the super-repo root directory. This file contains information about subprojects in this super-repo. Initially, we can just add the super-project to it:

     # The configuration file is empty for now.
     our-super-project-name {
       dir = "."
       origin = "<path-at-github-or-bitbucket>"
       mirrors = []
     }
    

    The syntax in the configuration file is HOCON. Alternatively, you can use JSON syntax for the project configuration file. Simply name the file repos.json and override the repositoriesFile method in your project definition to return repos.json. Then, add the empty JSON object inside that file:

     {
       "our-super-project-name": {
         "dir": ".",
         "origin": "<path-at-github-or-bitbucket>",
         "mirrors": []
       }
     }
    

    The repos.conf file is called the super-repo configuration file. At this point we can start SBT inside the super-repo.

Git Workflow

One of the main use cases for Mecha is simplified Git source control when using multiple projects. The requirement is that you have Git installed, and available in the PATH environment variable.

Optional Tracking

Let's start SBT inside mecha-super-repo. The first Mecha command that you need to know about is mecha-ls. This command will print all the subprojects from the super-repo configuration in repos.conf. If we run it for an empty configuration file, we get:

> mecha-ls
[info] Superproject repositories:
[success] Total time: 0 s, completed May 23, 2015 12:24:44 AM

So, let's add some projects. Assume that you have a project examples-core-utils, which contains core utilities. You would like to make changes its nightly version, but you want to use them right away (to verify that your changes work). The solution is to include examples-core-utils to our super-repo config in repos.conf (use your own fork for origin below, or create a fresh repo):

examples-core-utils {
  dir = "examples-core-utils"
  origin = "[email protected]:storm-enroute/examples-core-utils.git"
  mirrors = []
}

Alternatively, if you're using JSON syntax:

{
  "examples-core-utils": {
    "dir": "examples-core-utils",
    "origin": "[email protected]:storm-enroute/examples-core-utils.git",
    "mirrors": []
  }
}

Run reload in the SBT shell to load the change, then mecha-ls again:

> mecha-ls
[info] Superproject repositories:
[info] [ ] examples-core-utils
[success] Total time: 0 s, completed May 23, 2015 12:36:07 AM

Now we're talking business. The super-project is listing one registered subproject. If you inspect the directory structure in the super-repo, you will see that examples-core-utils is not really there. That is because you did not track it yet -- in general, there could be many projects registered, and you don't always want to check out all of them. So, let's track examples-core-utils -- for this we use the mecha-track command (auto-completion will list available projects):

> mecha-track examples-core-utils
[info] Cloning 'examples-core-utils' into 'C:\cygwin\home\...'.
Cloning into '.'...
remote: Counting objects: 3, done.
remote: Compressing objects: 100% (2/2), done.
remote: Total 3 (delta 0), reused 0 (delta 0), pack-reused 0
Receiving objects: 100% (3/3), done.
Checking connectivity... done.
[warn] Please reload the sbt shell.
[success] Total time: 4 s, completed May 23, 2015 12:39:29 AM

Yeehaw! The examples-core-utils project is now checked out. If we inspect the directory structure, we should be able to find it. Now reload the SBT shell one more time. Then, mecha-ls shows that examples-core-utils is checked out (note the *):

> mecha-ls
[info] Superproject repositories:
[info] [*] examples-core-utils
[success] Total time: 0 s, completed May 23, 2015 12:40:52 AM

Mecha will automatically add examples-core-utils to the .git/info/exclude file in the meta-repo so you don't accidentally commit it, unless it's already ignored via an existing pattern in either .gitignore or .git/info/exclude (e.g. you have examples-core-* in either of the files).
If examples-core-utils is whitelisted by an existing pattern (e.g. !examples-core-*), Mecha will log a warn, and you'll need to remove the conflicting pattern, then manually add examples-core-utils to the exclude file.
Unlike .gitignore, .git/info/exclude doesn't need to be committed to the meta-repo (it won't show up in untracked changes if you edit it).
For more info about gitignore, see the official documentation.

The examples-application project uses examples-core-utils for its benchmarks. Let's add examples-application to repos.conf too:

...
examples-application {
  dir = "examples-application"
  origin = "[email protected]:storm-enroute/examples-application.git"
  mirrors = []
}
...

And reload again -- mecha-ls now gives:

[info] Superproject repositories:
[info] [ ] examples-application
[info] [*] examples-core-utils

Do mecha-track for examples-application and reload once more, and now you're tracking both projects. Next, we will see how to pull from remote repositories.

Pulling

Pulling latest changes for all the projects is super-easy -- just do:

> mecha-pull
[info] Pull repo 'examples-core-utils' from origin...
Already up-to-date.
[info] Pull repo 'examples-application' from origin...
Already up-to-date.

And all the subprojects are updated. If some of the repos had pending changes, then the pull fails.

Committing

Let's make some changes. Add a line tmp/ to the .gitignore files in both examples-core-utils and the examples-application projects. Now run mecha-status:

> mecha-status
[info] ------ status for repo 'examples-core-utils' ------
[info] On branch master
[info] Your branch is up-to-date with 'origin/master'.
[info] Changes not staged for commit:
[info]   (use "git add <file>..." to update what will be committed)
[info]   (use "git checkout -- <file>..." to discard changes in working directory)
[info]
[info]  modified:   .gitignore
[info]
[info] no changes added to commit (use "git add" and/or "git commit -a")
[info] ------ end of status for repo 'examples-core-utils' ------
[info] ------ status for repo 'examples-application' ------
[info] On branch master
[info] Your branch is up-to-date with 'origin/master'.
[info] Changes not staged for commit:
[info]   (use "git add <file>..." to update what will be committed)
[info]   (use "git checkout -- <file>..." to discard changes in working directory)
[info]
[info]  modified:   .gitignore
[info]
[info] no changes added to commit (use "git add" and/or "git commit -a")
[info] ------ end of status for repo 'examples-application' ------

Looks like we've got some pending changes in both subprojects. We can also run 'mecha-diff' to precisely see the changes we made. If we are happy with the changes, we can run mecha-commit. This both stages the changes and commits them:

> mecha-commit
[info] --- diff for 'examples-core-utils' in 'examples-core-utils' ---
[info] diff --git a/.gitignore b/.gitignore
[info] index c58d83b..373b673 100644
[info] --- a/.gitignore
[info] +++ b/.gitignore
[info] @@ -15,3 +15,5 @@ project/plugins/project/
[info]  # Scala-IDE specific
[info]  .scala_dependencies
[info]  .worksheet
[info] +
[info] +tmp/
[info] --- end of diff for 'examples-core-utils' in 'examples-core-utils' ---
Commit message (empty aborts):

We can enter a message and the changes will be commited. If you are not happy with the output of the diff, just enter an empty message. We can do the same for the other repo -- Mecha will iterate through all the repos with pending changes. Once we're done, running mecha-status should report no pending changes.

Pushing

Once we committed the changes, we need to push them to origin. For this, we run mecha-push:

> mecha-push
[info] Push 'examples-core-utils' to 'origin'...
[info] Push 'examples-application' to 'origin'...
[info] ------ examples-core-utils ------
[info] To [email protected]:storm-enroute/examples-core-utils.git
[info]    cd260d4..fb527d5  master -> master
[info] ------ examples-application ------
[info] To [email protected]:storm-enroute/examples-application.git
[info]    edd27e1..5f6f5c6  master -> master
[success] Total time: 2 s, completed May 23, 2015 2:04:10 AM

And that's it!

Branching and Merging

Let's say that we want to work on feature gitignore. We would like to have a separate branch for that called topic/gitignore. To simultaneously create a new branch in all the repositories, run mecha-new-branch:

> mecha-new-branch topic/gitignore
[info] All repositories are at branch: master
Branch in all repos? [y/n] y
Switched to a new branch 'topic/gitignore'
Switched to a new branch 'topic/gitignore'
[success] Total time: 11 s, completed May 23, 2015 2:09:51 AM

Mecha will check if all the repositories are at the same branch, and ask you to confirm. If you do, the new branches are created. Let's check with mecha-branch:

> mecha-branch
[info] Repo 'examples-core-utils': topic/gitignore
[info] Repo 'examples-application': topic/gitignore

To switch all projects back to master, just run mecha-checkout:

> mecha-checkout
Existing branch name: master
Switched to branch 'master'
Your branch is up-to-date with 'origin/master'.
Switched to branch 'master'
Your branch is up-to-date with 'origin/master'.

To merge the changes from topic/gitignore back to master, run mecha-merge:

> mecha-merge
[info] All repositories are at branch: master
Merge in all repos? [y/n]y
Existing branch name (empty aborts): topic/gitignore
Already up-to-date.
Already up-to-date.
[success] Total time: 9 s, completed May 23, 2015 2:13:43 AM

You can alternatively specify the branch name directly in the command line.

Other Actions

You can additionally specify a list of mirrors in repos.conf for specific projects. If you do, you will be able to pull from and push to all the mirrors with mecha-pull-mirror and mecha-push-mirror.

You can also switch between projects with the project command. Since the subprojects are just normal SBT projects, you can work on them directly. Sometimes you need to do this to more easily invoked project-specific commands.

To switch between projects, you will need to specify their full path, since the subprojects are defined as sbt RootProjects.

project {file:/C:/cygwin/home/scala-fanboy/workspaces/scala/examples/mecha-super-repo/examples-core-utils}

Luckily, auto-complete's here to the rescue!

Project Configuration and the Input Query DSL

In some projects, it is useful to have configuration files that the users need to fill out before being able to build the project. Examples of values in this configuration file include sensitive information such as passwords, local paths or other dev-specific info that you do not want in source control. The common convention is to provide a template configuration that the devs must fill. Sadly, the configuration template and the actual configuration can easily get out-of-sync and the devs must remember to fill them out correctly each time.

Mecha has an Input.Query DSL that you can use to define these configuration files in the subprojects. This DSL automatically creates a configuration file generator that queries the user to enter values for the config file (if the config file does not exist). Let's say that examples-application needs to have a different target location and a version version. We first need to convert the examples-application build into a Mecha repo build.

  1. Add Mecha to project/plugins.sbt in examples-application:

     resolvers ++= Seq(
       "Sonatype OSS Snapshots" at
         "https://oss.sonatype.org/content/repositories/snapshots",
       "Sonatype OSS Releases" at
         "https://oss.sonatype.org/content/repositories/releases"
     )
     
     addSbtPlugin("com.storm-enroute" % "mecha" % "0.2")
    

    If necessary, replace 0.2 with the latest Mecha version.

  2. Add the following into the project/Build.scala:

     import org.stormenroute.mecha._
     import sbt._
     import sbt.Keys._
     
     object ExamplesApplicationBuild extends MechaRepoBuild {
    
       lazy val examplesApplicationSettings = MechaRepoPlugin.defaultSettings ++ Seq(
         name := "examples-application",
         scalaVersion := "2.12.1",
         version := "0.1",
         organization := "com.storm-enroute",
         libraryDependencies ++=
           superRepoDependencies("examples-application"),
       )
     
       def repoName = "examples-application"
    
       lazy val examplesApplication: Project = Project(
         "examples-application",
         file(".")
       ).settings(
         examplesApplicationSettings
       ).dependsOnSuperRepo
     }
    

    Here, the crucial part is the dependsOnSuperRepo -- don't forget to add this or the dependencies won't be picked up! The other crucial part is libraryDependencies ++= superRepoDependencies. For repoName, use the same name as in the repos.conf file from the super-repo.

  3. Reference build definition from build.sbt as in the following example:

     lazy val `examples-application` = ExamplesApplicationBuild.examplesApplication
    
  4. Run reload in the SBT shell and you've got a Mecha repo build.

The examples-application can now do various powerful stuff. Let's get back to our config files. We add the following value to project/Build.scala in examples-application:

import Input._
val configQuery = {
  val target =
    const("file(\"/custom-target/\")")
      .map(("target", _))
  val version =
    stringQuery("Enter version: ")
      .map(v => "\"" + v + "\"")
      .map(("version", _))
  traverseFull(List(target, version))
}

The above is a template for generating a configuration. The const value will default its argument. For each stringQuery value, the user will have to enter the value when SBT boots. You can add more vals if you need more configuration values.

Importantly, add the following setting to examplesApplicationSettings:

...
MechaRepoPlugin.configQueryKey := Some(configQuery),
...

Now reload the project, run package and see the magic happen:

> package
[warn] Populating configuration file.
Enter version: 1.7.4
[error] Created 'C:\cygwin\home\...\config.sbt'. Please reload!

The generated file is called config.sbt. You can change the default name with the MechaRepoPlugin.configFilePathKey setting.

Whaa? How does this configuration input-query DSL work?

There is no need to understand this DSL to use it. However, if you really want to know, read this section.

The Input.Query DSL relies heavily functional combinators to create queries and generation values. It is based on the following type:

type Query[T] = () => Option[T]

This reads "A query of type T is code that maybe produces a value of type T". When we call const(x), we produce a function that always returns Some(x):

def const[T](v: =>T): Query[T] = () => Some(v)

The stringQuery asks the user to produce a query -- it might return None if the user enters an empty string. Both const and stringQuery are basic combinators.

Complex combinators transform existing Query objects into more complex ones. For example, the map that we used above transforms the value in the Option object that Query could return if the Option is not None. We used it to wrap the user's string into quotes.

Another example is repeat -- given a Query it returns another Query that repeats the original query until the user enters an empty string. Other combinators include default, map, traverse, traverseFull, pair, ...

The configQueryKey is a setting for values of the following type:

Option[Query[Traversable[(String, String)]]]

This reads: an optional query that may return a traversable of string tuples. These tuples are the key-value pairs that end up in the configuration objects.

Inter-Project Dependencies

Now the main reason why we have Mecha -- projects can depend on each other, and sometimes you don't want to wait until the snapshot lands on Maven to get the updates from one project in another project.

In these cases, you define a dependency file inside your subproject. The dependency file in project A can, for example, specify that project A depends on some other project B. If you run SBT from the superproject root directory, Mecha will try to create a source code dependency on project B for project A. If the project B is not tracked, Mecha will create a dependency on the Maven artifact instead. Similarly, if you checkout project A without the superproject, Mecha will create a regular Maven dependency.

In our example, examples-application project will depend on the examples-core-utils project. To express this dependency, we need to:

  1. First remove the dependency on the other project from the libraryDependencies setting, if there is one.

  2. Next, create a dependencies.conf file in the root directory of the examples-application project:

     examples-application {
       project = "examples-core-utils"
       artifact = ["com.storm-enroute", "examples-core-utils", "0.1"]
     }
    

    Alternatively, you can use JSON syntax, and a dependencies.json file. Simply override both dependenciesPath and repositoriesPath fields in the project definition, and add the following to the dependencies.json file:

     {
       "examples-application": [
         {
           "project": "examples-core-utils",
           "artifact": ["com.storm-enroute", "examples-core-utils", "0.1"]
         }
       ]
     }
    

Here, for every SBT project inside the examples-application subproject, we specify a list if dependent projects. In case the dependent projects are not checked out, we fall back to their published artifacts.

We can also specify that the dependency is only on a specific project configuration:

artifact = ["com.storm-enroute", "examples-core-utils", "0.1", "test"]

Now we can e.g. call methods from examples-core-utils directly from examples-application. Assume that we have the following in examples-core-utils:

package com.stormenroute

object Util {
  def version = println("core-utils v1.0")
}

We add the file src/main/scala/com/stormenroute/Main.scala to examples-application:

package com.stormenroute

object Main {
  def main(args: Array[String]) {
    println(Util.version)
  }
}

And hit compile.

Specifying Inter-Project Dependencies in the Source Code

If you don't like using separate dependencies.conf file for specifying dependencies, you now have an option to specify dependencies in the source code of your build, and have all the dependencies in one place.

Here is an example project build setup, in project/Build.scala:

    import org.stormenroute.mecha._
    import sbt._
    import sbt.Keys._

    object ExamplesApplicationBuild extends MechaProjectBuild {

      lazy val examplesApplicationSettings = MechaRepoPlugin.defaultSettings ++ Seq(
        name := "examples-application",
        scalaVersion := "2.12.1",
        version := "0.1",
        organization := "com.storm-enroute",

        // include dependencies as usual
        libraryDependencies ++= runtimeDependencies.value,

        // AT THE END of settings
        // exclude dependencies based on super-repo, if any
        libraryDependencies --= excludeSuperRepoDependencies.value
      )

      // put super-repo projects dependencies here
      val superRepoProjectsDependencies: Seq[(String, String, Option[String])] = Seq(
        // (repo, project, configuration)
        ("examples-core-utils", "examples-core-utils", None)
      )

      // you can put ALL usual project dependencies here
      val runtimeDependencies: Def.Initialize[Seq[ModuleID]] = Def.setting(Seq(
        Def.setting("com.storm-enroute" %% "examples-core-utils" % "0.1").value,
        Def.setting("org.scalatest" %% "scalatest" % "3.0.1" % "test").value
      ))

      val repoName = "examples-application"

      lazy val examplesApplication: Project = Project(
        "examples-application",
        file(".")
      ).settings(
        examplesApplicationSettings
      ).dependsOnSuperRepo
    }

Here, the crucial parts are:

  • dependsOnSuperRepo don't forget to add this or the dependencies won't be picked up!
  • libraryDependencies --= excludeSuperRepoDependencies.value add this AT THE END of project settings to exclude dependencies based on super-repo
  • For repoName, use the same name as in the repos.conf file from the super-repo
  • Instead of using dependencies.conf file you can now specify project dependencies inside superRepoProjectsDependencies and runtimeDependencies lists, they are described bellow

Using superRepoProjectsDependencies you can specify a list of source-level dependencies on projects your project depend on inside the super-repo.

And using runtimeDependencies you can specify a list of project runtime dependencies artifacts as usual, as if there is no super-repo, or repository is not tracked.

We have to wrap dependencies inside runtimeDependencies into Def.setting(...) because of a value sbt macro, which is implicitly used when some cross-project dependencies are specified with the %%% macro, which in turn can only be used inside a setting definition:

      val superRepoProjectsDependencies: Seq[(String, String, Option[String])] = Seq(
        // repo, project, configuration
        ("examples-core-utils", "examples-core-utils", None),
        ("test-utils", "some-more-utils", Some("test")),
        ("ui-repo", "some-js-project", None)
      )

      val runtimeDependencies: Def.Initialize[Seq[ModuleID]] = Def.setting(Seq(
        Def.setting("com.storm-enroute" %% "examples-core-utils" % "0.1").value,
        Def.setting("com.storm-enroute" %% "some-more-utils" % "0.1" % "test").value,
        Def.setting("com.storm-enroute" %%% "some-js-project" % "0.1").value
      ))

Edit-Refresh Task

Mecha comes with a special mechaEditRefresh task that supports fast edit-refresh cycles across projects. Essentially, calling this task in the super-repo depends on completing that task in all the subprojects. For specific Mecha subprojects, make sure that this task is overridden, for example:

mechaEditRefreshKey <<= mechaEditRefreshKey.dependsOn(compile in Compile)

Then, in SBT shell do:

> ~mecha-edit-refresh

And code the night away.

Automated Docs Publishing

Mecha supports automated publishing of your project ScalaDoc to a target Git repo. To enable this, you need to specify these settings:

...
mechaDocsRepoKey := "url of the git repo for hosting the docs",
mechaDocsBranchKey := "branch in the remote git repo",
mechaDocsPathKey := "the path in the repo where the docs should be put",
...

Running the mecha-publish-docs command from the subproject will:

  1. Build the docs.
  2. Check out the specified branch of the Git url into a temporary directory.
  3. Copy the docs into the path specified in mechaDocsPathKey, suffixed with version.
  4. Amend the last commit and force-push back to the remote Git repo.

The fun part with this command is that if you use GitHub and specify a gh-pages branch, your docs will be automatically hosted for the whole world to see... So, you get ScalaDoc hosting for free.

Automated Benchmark Publishing

Automated benchmark publishing is very similar to docs publishing, and is triggered with mecha-publish-benches in the subproject. You will need to specify these settings:

...
mechaBenchRepoKey := "url of the git repo",
mechaBenchBranchKey := "branch in the git repo",
mechaBenchPathKey := "subdirectory in the git repo",
mechaBenchSrcPathKey := (baseDirectory.value / "target/benchmarks").toString,
...

See descriptions of these settings for more info.

Automated Build Deployment

Automated build deployment is similar to docs and benchmark publishing, and is triggered with mecha-publish-build in the subproject. The difference is that each build is deployed to a subdirectory with a timestamp and SHA-1 has of the commit. You will need to specify these settings:

...
mechaBuildOutputRepoKey := "url of the git repo where builds are copied",
mechaBuildOutputBranchKey := "branch in the git repo",
mechaBuildOutputPathKey := "subdirectory in the git repo",
mechaBuildOutputSrcPathKey := (baseDirectory.value / "target").toString,
mechaBuildOutputExpirationDaysKey := 5, // number of days builds before removing old builds
...

See the descriptions of these settings for more info.

Nightly Task

Mecha has another special task called mechaNightly. By default, this task depends on packaging, testing, publishing the project, publishing the docs and publishing the benchmarks. If you want to extend this, add more dependencies for this task inside the subproject.

Calling mecha-nightly in the super-repo depends on all the Mecha subprojects.

Automated Remote Deployment

In case you're working on a project that needs to be frequently deployed to a different host during development, Mecha exposes the mechaSshDeploy command to speed this up. You will need to specify the following settings:

...
remoteSshHost := None,
remoteSshUser := "admin",
remoteSshPass := None,
remoteDeployPathKey := "~",
remoteDeployCmdKey := None,
...

Calling mecha-ssh-deploy from within the subproject will:

  1. Connect to the remote host, using the specified credentials.
  2. Pull the project into the specified path.
  3. Optionally run the bash command from remoteDeployCmdKey in the project root.

Note: commit your changes before deploying them at the remote host -- the project is pulled at the remote location, but there is no local automatic commit/push.

Some of these settings, like remoteSshPass are ideal candidates for the Input Query DSL and the automated project configuration.

About

SBT workflow plugin for development, testing and publishing of multiple project repositories simultaneously.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published