Deciding on a strategy to optimize your software is a complex and delicate task that depends on the application type, target platforms, and business requirements. In this chapter, we provided some guidelines to help you think and choose an appropriate software stack for your own applications.

High-performance numerical applications sometimes require managing installation and deployment of third-party packages that may require handling of external tools and native extensions. In this chapter, we saw how to structure your Python project, including tests, benchmarks, documentation, Cython modules, and C extensions. Also, we introduced the continuous integration service Travis CI, which can be used to enable continuous testing for your projects hosted on GitHub.

Finally, we also learned about virtual environments and docker containers that can be used to test applications in isolation and to greatly simplify deployments and ensure that multiple developers have access to the same platform.