Need for some way to test the oxygen-tei package before release

[TEITechnicalCouncil]

Since the new plugin update system in Oxygen means that people are likely to update their TEI frameworks in Oxygen quite rapidly, it would be very useful to have a small test suite that somehow checks that the latest integration of a P5 release and a Stylesheets release actually works OK before unleashing it on the world. I don't know exactly what we would test or how yet, though...

Original comment by: @martindholmes

[TEITechnicalCouncil]

This issue was originally assigned to SF user: martindholmes Current user is: martindholmes

[TEITechnicalCouncil]

• Description has changed:

Diff:

--- old
+++ new
@@ -0,0 +1 @@
+Since the new plugin update system in Oxygen means that people are likely to update their TEI frameworks in Oxygen quite rapidly, it would be very useful to have a small test suite that somehow checks that the latest integration of a P5 release and a Stylesheets release actually works OK before unleashing it on the world. I don't know exactly what we would test or how yet, though...

Original comment by: @martindholmes

[TEITechnicalCouncil]

• assigned_to: Martin Holmes
• Group: AMBER --> RED

Original comment by: @jamescummings

[TEITechnicalCouncil]

Putting Martin as owner for now. Seems relatively uncontroversial but changing to RED since we need at least some guess at how to do this. (As a side note we need to document the oxygen-tei auto-update and how to set it up more clearly.)

Original comment by: @jamescummings

[TEITechnicalCouncil]

Two suggestion arose during discussion at Council meeting: 1) make backing out the framework that is auto-updated a very easy process 2) have a beta version available where we put the framework a week before real release, so we can test it ourselves. (Just suggesting that we test it basically work, not test every feature.)

Original comment by: @sydb

[TEITechnicalCouncil]

• Group: RED --> AMBER

Original comment by: @martindholmes

[TEITechnicalCouncil]

Setting this to amber now that https://sourceforge.net/p/tei/feature-requests/526/ is complete. We still need to discuss what form the tests should take; I imagine an Oxygen project file with a bunch of ant tasks which create files from templates, do transformations, validate etc., which people could run manually, but we could probably also create some automated tests which run as part of the Jenkins job.

Original comment by: @martindholmes

[sydb]

Council sub-group thinks having tests is a good idea, and this ticket should be GREEN for @martindholmes to go ahead. We also think that this is probably a lot of work, and is not a particularly high priority. Lastly, we think that our thoughts of 2014-07-02 still hold.

[ebeshero]

@martindholmes Any progress here?

[martindholmes]

Sorry, overwhelmed with other work. I'd like to get someone on Council who knows Oxygen well (Magdalena???) to help specify what the most appropriate set of tests would be. We have the mechanism in place for doing it (the plugin bleeding-edge build job) but figuring out what would make a useful but not excessive set of tests is not trivial.

[martindholmes]

@tuurma With your expert knowledge of Oxygen, what would be your recommendation for an effective way to program tests for the framework?

[emylonas]

not assigning because not sure of the status of this ticket. It's old, hasn't been discussed since 2016, and @martindholmes is divorcing himself from it. Is it the discussion still relevant or have we moved on in our thinking about the Oxygen plugin and running tests for the framework? Still seems low priority, so perhaps best to leave unassigned.

[raffazizzi]

Given the lack of clarity, I'm relabeling as "Needs Discussion", but leaving "Go" to remind us that this was approved at some point. If this ticket remains dormant until Spring 2020 F2F, please discuss then.

[martinascholger]

@hcayless @martindholmes @tuurma I think we can close this for now?

[martindholmes]

@martinascholger I think nobody is ever going to do anything about it. But we just demonstrated the need for it conclusively last week: we "released" a plugin upgrade that wasn't there.