You may be able to avoid this initial setup if the build job is copied from an
existing build. The genisis build system is edinburgh_buildtest. Look there for
examples.
1. You must create a job for each build stream entitled
dependency_calculator.<build_name>
All the groovy scripts and the dependencies.properties file must be copied
here. The idea is that there is a single place to make updates to the
dependencies as well as to fix the scripts if they need. Once a change is
made to any one of these files the dependency_calculator.<build_name> must
be run so that the changes are propagated to the reset of the build.
2. The dependency.properties file will have to modified as the dependencies
change with each build. Currently this is a manual process but it is hoped
that we will be able to automate this in the future.
3. The dependencies are fairly straightforward represenations of what appears in
either the maven dependencies or Jenkins copy artifact. The exceptions are
as follows.
A. Dependencies that arise because of copying artifacts directly in the pom
files. This has been taken into account in the current dependencies file
but will have to modified if we want to automate the process.
B. The parent pom files will have dependencies on any job that the children
are dependent that are not in the source tree for the parent pom. The
reason that this was done is that a build will not continue if there
are no changes recognized in the parent pom job. We must account for
lower level jobs that are dependent on external jobs.
4. All jobs that are calculated as not needed will be marked as ABORTED with a
wrapper after the job to compensate for this status. As will all things there
is an exception. When the parent job is calculated as not needed it will set
the status to NOT_BUILT which in turn will cause the children jobs to not
execute. That is why there are some jobs that have an fix_not_built_status
as opposed to fix_abort_status wrapper.
5. As much as possible I tried to encapsulate the groovy jobs that fix status
with the jobs for which may have been calculated as not needed. Also the
build_avoidance.groovy file is encapsulated with the job that copies the
script into the current work space. It's just a logical construct and
will aid when we have to rearrange jobs.
6. The build_conversion_notes.txt is not complete but I'm including it here
in the hopes that I can update it at the end of the build conversion
process. I also intend to turn this content into a wiki page for
future reference.