Release 0.27

• ENHANCEMENT

  The new create-jenkins-user command creates a user account in a Jenkins installation to which filesystem-level access is available.

• ENHANCEMENT

  The new install-jenkins command downloads, installs and configures a Jenkins instance into a given directory. To avoid Jenkins' initial setup wizard, a user account is also created.

• ENHANCEMENT

  When analyzing Git repositories, the git ls-remote invocation is no longer mandatory in all cases. Instead, results of these invocations are stored in a timestamped cached and reused while still recent enough. The new
  configuration option global.cache-age-limit and the corresponding commandline option cache-age-limit control how long after their creation cached results are still considered recent enough. The default value is 1800 seconds.

• ENHANCEMENT

  The new config command allows inspecting configuration sources and the current configuration.

Release 0.26

• ENHANCEMENT The HTML report publisher plugin for Jenkins is now supported.

• ENHANCEMENT When the first attempt at cloning a mercurial repository fails, a second attempt without using "bundles" is made.

• ENHANCEMENT Error reports printed during the analysis and deployment phases now always include the project and/or project version name.

• ENHANCEMENT The "Jenkins Warnings Plugin - Next Generation" plugin is now supported.

  However, generated job configurations continue to select the legacy warnings and tasks plugins by default. The following commandline options can be used to generate job configurations which select the new plugin:

  -D aspect.checkstyle.implementation=ng
  -D aspect.pmd.implementation=ng
  -D aspect.tasks.implementation=ng
  -D aspect.warnings.implementation=ng

• ENHANCEMENT The behavior in case of errors can now be controlled at a finer level of granularity. To this end, the on-error commandline option, in addition to the previous values, now accepts a policy specification adhering to the following grammar:

  error-policy ::= rule* default
  rule         ::= error "=>" action ":"
  error        ::= "object-error" | "simple-object-error"
                   | "syntax-error" | "repository-access-error"
                   | "repository-analysis-error"
                   | "project-analysis-error" | "analysis-error"
                   | "dependency-error" | "instantiation-error"
                   | "deployment-error" | "report-error"
  default      ::= action
  action       ::= "abort" | "fail" | "continue" | "debug"
  

  Example:

    dependency-error=>continue:analysis-error=>fail:abort
  

  The above continues the run with exit code zero in case dependency-errors are encountered, continues and returns a non-zero exit code for analysis-errors and immediately aborts with non-zero exit code for all other errors.

• ENHANCEMENT Distribution recipes can now include other distribution recipes in a new include section. Distribution includes can be written in two forms, either

  include:
  - NAME

  or

  - name: NAME
    parameters:
      PARAMETER₁: VALUE₁
      PARAMETER₂: VALUE₂
      ⋮

  where NAME is the name of a distribution (without file extension) and PARAMTERₖ: VALUEₖ are variable bindings that should overwrite bindings in the included distribution. Included distributions can in turn include other distributions. Variables referenced in projects (and jobs and aspects) are now looked up in the following order (most specific, highest priority first):

  • Specified as a parameter when (directly) including the project version in a distribution
  • Specified as a variable in an entry in the versions section of the project recipe
  • Specified as a variable in the global variables section of the project recipe
  • Specified as a parameter when including the distribution which in turn directly includes the project version
  • Specified as a variable in the distribution which directly includes the project version
  • Specified a as variable in the "outermost" (that is, given on the commandline) distribution
    If a given project version is included in multiple distributions, the instance (and in particular the parameters of that instance) that appears first in terms of a depth-first traversal of include relations is used while
    all others are discarded. Note that the textual order of the include and versions sections in distribution recipes has no bearing on the ordering described above.