After several conversations with folks, I don't think that this message
has come through very clearly so I want to re-emphasize something that
is a core design philosophy for Taskotron:
You do not need a full Taskotron system deployment to run tasks.
Yes, the production system consists of a set of services to handle job
triggering, reporting and other support services. However, the core
running of tasks can be done with a git checkout of libtaskotron.
Following the installation instructions in the readme:
Unless you're doing integration testing of the various production
components or working on the integration code between components like
libtaskotron and buildbot or trigger, you should be able to work on
tasks without _anything_ beyond libtaskotron and its direct
dependencies. If you ever find this not to be the case, let someone
know - it's almost definitely a bug.
The ability to run tasks without a full production-ish deployment is an
important feature for Taskotron. It's why we took such a large step
backwards instead of trying to directly "fix" AutoQA and why the
complexity inherent to an automation system has been moved around. It
_needs_ to be relatively trivial for non-core contributors to run and
develop tasks for Taskotron.
Taskotron as a whole is not going to be successful unless we can get
contributions from other groups/people and that's not likely to happen
unless task development and maintenance is as easy as we can possibly