Details
-
Type:
Improvement
-
Status: To Do
-
Resolution: Unresolved
-
Fix Version/s: None
-
Component/s: Continuous Integration
-
Labels:None
-
Team:SQuaRE
Description
Several packages in the stack include example code which is neither included as part of the build nor exercised in the unit tests. See, for example afw and pipe_tasks. This means it's all too easy for folks to forget to update these examples when they make changes elsewhere. It would be great if we could execute the examples and demonstrate that they still work as part of the CI system.
Attachments
Issue Links
- blocks
-
DM-9389 Provide a working IsrTask example
- Won't Fix
-
- Won't Fix
- relates to
-
DM-9142 calibrateTask.py example not working on current stack
- Won't Fix
-
- Done
- mentioned in
-
Page Loading...
Activity
I would think that all of those vary on a case-by-case basis.
I guess it's not clear from the above, so: it's not my suggestion that we should be able to flip a switch and have all of our existing examples run through CI (if nothing else, the resulting breakage would be a huge timesink). Rather, I'd request that SQuaRE take the lead in defining a way of creating "executable examples" (of a variety of types – should include IPython notebooks, if possible) in such a way that they will be CI-able in future.
CI-able examples are definitely part of the documentatioin plan.
Just updating this comment thread of Oct 2015 with my tentative answer to Tim Jenness above – my take would be that yes, any example that's publically provided be runnable, and working in the current version of the Stack, and included in CI as well.
Don't have a timescale for when all of this should fully be implemented, but I do think it should definitely be the overall goal, if everyone agrees.
Reviewed for DM-CCB, 2020-07-15. We continue to have no way to CI our examples, and it continues to be a major problem.
Are all examples expected to be runnable? What is the measure of success? Do the examples know how to find their own test data?