35.0 Mr. Robot¶
This is another proud announcement: Avocado release 35.0, aka “Mr. Robot”, is now out!
This release, while a “regular” release, will also serve as a beta for our first “long term stability” (aka “lts”) release. That means that the next release, will be version “36.0lts” and will receive only bug fixes and minor improvements. So, expect release 35.0 to be pretty much like “36.0lts” feature-wise. New features will make into the “37.0” release, to be released after “36.0lts”. Read more about the details on the specific RFC.
The main changes in Avocado for this release are:
- A big round of fixes and on machine readable output formats, such as xunit (aka JUnit) and JSON. The xunit output, for instance, now includes tests with schema checking. This should make sure interoperability is even better on this release.
- Much more robust handling of test references, aka test URLs. Avocado now properly handles very long test references, and also test references with non-ascii characters.
- The avocado command line application now provides richer exit status. If your application or custom script depends on the avocado exit status code, you should be fine as avocado still returns zero for success and non-zero for errors. On error conditions, though, the exit status code are richer and made of combinable (ORable) codes. This way it’s possible to detect that, say, both a test failure and a job timeout occurred in a single execution.
- [SECURITY RELATED] The remote execution of tests (including in Virtual Machines) now allows for proper checks of host keys. Without these checks, avocado is susceptible to a man-in-the-middle attack, by connecting and sending credentials to the wrong machine. This check is disabled by default, because users depend on this behavior when using machines without any prior knowledge such as cloud based virtual machines. Also, a bug in the underlying SSH library may prevent existing keys to be used if these are in ECDSA format. There’s an automated check in place to check for the resolution of the third party library bug. Expect this feature to be enabled by default in the upcoming releases.
- Pre/Post Job hooks. Avocado now defines a proper interface for extension/plugin writers to execute actions while a Job is runnning. Both Pre and Post hooks have access to the Job state (actually, the complete Job instance). Pre job hooks are called before tests are run, and post job hooks are called at the very end of the job (after tests would have usually finished executing).
- Pre/Post job scripts. As a feature built on top of the Pre/Post job hooks described earlier, it’s now possible to put executable scripts in a configurable location, such as /etc/avocado/scripts/job/pre.d and have them called by Avocado before the execution of tests. The executed scripts will receive some information about the job via environment variables.
- The implementation of proper Test-IDs in the test result directory.
Also, while not everything is (yet) translated into code, this release saw various and major RFCs, which are definitely shaping the future of Avocado. Among those:
- Introduce proper test IDs
- Pre/Post test hooks
- Multi-stream tests
- Avocado maintainability and integration with avocado-vt
- Improvements to job status (completely implemented)
For a complete list of changes please check the Avocado changelog. For Avocado-VT, please check the full Avocado-VT changelog.
Instructions are available in our documentation on how to install either with packages or from source.
Updated RPM packages are be available in the project repos for Fedora 22, Fedora 23, EPEL 6 and EPEL 7.
As a heads up, we still package the latest version of the various Avocado sub projects, such as the very popular Avocado-VT and the pretty much experimental Avocado-Virt and Avocado-Server projects.
For the upcoming releases, there will be changes in our package offers, with a greater focus on long term stability packages for Avocado. Other packages may still be offered as a convenience, or may see a change of ownership. All in the best interest of our users. If you have any concerns or questions, please let us know.
Happy hacking and testing!