Test Program Generator
Documentation of a test program is important to allow non-test-engineering stakeholders to understand what is in a given program, or what a given bin number refers to. So Origen fully embraces the notion that test programs should be documented, that this documentation should not have an opportunity to be wrong, and that no one should have to waste their time manually trying to keep a document and the program in sync.
However many tools in this area approach this problem by trying to turn documentation into a test program. Origen turns this problem on its head and unashamedly asserts that the test program itself is the true source of authority as to what it does, and that documentation is the follower instead of the leader.
This approach has the following advantages:
- Test programs are large and complex, who really wants to write the whole thing out manually? It is much more efficient (and less error prone) to write it in code, then we can use subroutines, variables, object-orientation, meta-programming (code that generates other code) and any other programming techniques to eliminate duplication and opportunities for error.
- Document first approaches are always constrained by the fact that some tests are more complicated than others. In a recent industry conference a leading document-first generator stated that typically it could handle 80% of tests, for the other 20% you were on your own. With Origen’s approach 100% of your tests will be generated.
One problem with the Origen approach is how do you bootstrap a new program, intuitively you may want to document it first.
This problem can be solved via the following workflow:
- The Flow file syntax is designed to be as simple and human-readable as possible, and it is self contained and can exist long before any interfaces are ever written. Therefore it you know enough to be able to write a document of a flow, then you know enough to be able to write an Origen flow file for that flow.
- Nonetheless some early stakeholders may want to get involved in the definition but won’t feel comfortable with the code layer. In this case it is recommended that you immediately setup a documentation interface (these are trivially simple and will probably eventually come from free with Origen), and begin generating documentation of the flow from day 1.
- Definition can continue as required by iterating on the flow file, all the while the documentation stays in sync.
- When it comes to finally implementing the flow the test engineering team will create the interface which will turn it into a real test program.
An alternative approach is to have a short-lived hand-written document that is used to get the team up to the first release of the test program. At that point authority should switch to the program itself and a ticketing system should be used to iterate on the program…
The underlying program application should use a ticketing system. Any non-test-engineering stake holders can review the documentation to see what the test program currently does, then if they want it to do something else they will raise a ticket requesting it.
Once the change has been implemented the ticket can be closed and the documentation will automatically pick up the change.
This workflow is much easier to monitor and track status vs. the alternative approach where the non-technical stakeholder release new versions of a document and the TE has the job of trying to keep the program in sync.
As the flow file is written in Ruby, it is perfectly feasible to build higher level structures such as an importer, or even a spreadsheet-based interface on top of this.
The early users of the Origen test program generator have not had a need for such an approach, however this is definitely an area where the community could start to build new features.
It’s Origen’s job to provide a solid generation platform, however it deliberately does not try to dictate how a functional/parametric/bitmap/ATD ramp/etc. test should be set up. Firstly the scope of such an effort is too large, but secondly the Origen developers do not necessarily have the knowledge required to dictate what is the best practice in a lot of these areas.
However the idea of eliminating the interface layer is appealing and the good news is that it is achievable.
The Origen Plugins feature provides a way for Origen code to be easily packaged into a re-usable plugin and shared between projects. So for example, an ATD interface module could be written and then anyone wishing to create ATD tests, could import this into their application and completely bypass the interface creation step altogether. Just call the interface methods from the flow file and you will get a guaranteed working (and hopefully efficient!) ATD test.
If you are a test expert in a given area, and you are using Origen to generate your program, then we would encourage you to share your expertize by making your test setups available as an Origen plugin.