Andreas Platschek, Nicholas Mc Guire, Georg Schiesser, Thomas Hisch, OpenTech EDV Research GmbH, Austria

Writing code for safety critical systems implies a lot of restrictions and requirements in the software development process. Depending on the field a specific development process has to be followed and fully documented. In order to be compliant with a standard ,e.g. DO-178B, ISO61508, ISO26262, MISRA-C, stringent traceability requirements must be followed allowing to justify the development process itself.

In this paper, we try to give some examples of features of git that are - in our eyes - useful in helping the developers to fulfill the traceability and documentation requirements of safety related development life-cycles. Examples for such techniques are a history for each and every line of code, or the sign-off mechanism, to find out who is responsible for any part of the code base - providing a method to ensure you know who to blame.

These git specific features can even be improved by using git hooks in combination with different tools, including formal methods, and back-end scripting, allowing to fully automate these QA related extensions.

The intention behind all of these techniques is to build a tool-chain for extended traceability [version control] around git, simplifying the verification process.
Although this paper targets safety critical systems in particular, some of the presented techniques may well be suited for enhancing the kernel development as well as for standard user space programing tasks.

[DISCLAIMER:] None of the proposed techniques replaces common sense, they are just a way of improving your code and development process.