Testing 1.2.0 on ci

[jspaaks]

Related issues

Refs: #none

(For autoclosure of issues when PR is merged use Fixes #<issue-number> syntax)

Describe the changes made in this pull request

This PR

• fixes some merge errors in the schema
• changed the test collection setup to limit itself to tests/1.0.3 and tests/1.1.0
• adds tests/1.2.0/poc with a proof of concept data file
• adds testing of the poc data file as part of the CI

Instructions to review the pull request

GitHubh Action should run tests for 1.0.3 and 1.1.0 data, then run the poc schema validation in a separate step

If I add some random keyname as part of an author, the test breaks (https://github.com/citation-file-format/citation-file-format/pull/196/checks?check_run_id=2720914662), albeit in an unexpected way (it complains about date-start for example). Not sure if that's something we should address.

[sdruskat]

If I add some random keyname as part of an author, the test breaks (https://github.com/citation-file-format/citation-file-format/pull/196/checks?check_run_id=2720914662), albeit in an unexpected way (it complains about date-start for example). Not sure if that's something we should address.

Interesting. I've had this before, if I remember correctly in cases where we have different combinations of required/non-required keys or checking against different patterns. I think in these cases jsonschema tries to (heuristically?) guess the target object type, but not reliably so.

In the case you link to I think it breaks due to the injected bad key (which is correct), but also tries to match against person (instead of entity), and so lists all of the keys it wouldn't expect for person. Could be a bug in jsonschema.

EDIT: Reported as python-jsonschema/jsonschema#812.

[sdruskat]

Thanks! Good to merge :)

[jspaaks]

Awesomesauce 🍪 🍰