Merge ~ahasenack/qa-regression-testing:qat-openssh-fixes-and-fido-support into qa-regression-testing:master
Status: | Merged | ||||
---|---|---|---|---|---|
Merged at revision: | 00699043c116b02e6f60ecef2e5289c7caff907d | ||||
Proposed branch: | ~ahasenack/qa-regression-testing:qat-openssh-fixes-and-fido-support | ||||
Merge into: | qa-regression-testing:master | ||||
Diff against target: |
396 lines (+309/-8) 2 files modified
notes_testing/openssh/README.txt (+165/-0) scripts/test-openssh.py (+144/-8) |
||||
Related bugs: |
|
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Steve Beattie | Approve | ||
Review via email: mp+381582@code.launchpad.net |
Description of the change
Tests for openssh with U2F support.
This is the final requirement for the libfido/libcbor MIR[1], as agreed to in [2].
Instructions were added to notes_testing/
I was just able to add basically 2 automated tests. One generates the key (and is not a unit test, because I need that key for the next test), and the other uses it for a ssh localhost check. I would have loved to add tests to verify ssh fails if the device isn't present, but all this interactivity doesn't work well with the unittest framework. I would have to keep asking the user stuff like "Please remove the hw key now", then check that it was removed, then move on to trying ssh, and then ask for the key to be put back it for the next test in the suite. I accept suggestions here if you think this should be improved.
That being said, manual testing isn't that hard, and I also contemplated that in the README.txt document.
Please let me know what you think, or if something is too confusing and needs clarification, or could be done in a better way.
Note I also had to fix the openssh tests, as a few were expecting a password-based authentication. These changes are in separate commits.
1. https:/
2. https:/
Thanks Andreas, looks good to me!