Enhancing the functionality of the reproducibility script
Fixes: #1625
We have some enhancements that we would like to introduce into the test_reproducibility.py
script.
- We would like to add a parameter to enable
negative
test cases regarding the reproducibility script and ensure the behaviour is aligned with what is expected. This is a schedule that I have set up on a forked instance. Here we have twohello-world
examples being run. One uses the flag, and one does not. Both are markednon-reproducible
, yet one passes because we told the script that expected the job to fail, i.e., marked asnon-reproducible
. Hence it still returns an exit code of 0 in this instance. The schedules fulfil and provide evidence for theacceptance criteria
as laid out in #1625.
Please note that the schedules point to a public repository that has the infrastructure needed for me to run CI pipelines. This MR instance contains only the update to the script itself
-
We would like to update the final message to inform developers to look for a
HTML
pagel not a.txt
file. -
We would like to serialise the final output of each element at the end of a pipeline and provide it as an additional artifact. The HTML page is great; however, we should use a widely-applicable standard (like JSON) to store the results. JSON will enable us to send the data to other applications for consumption and post-processing.
CC:
Edited by Tanvir Roshid