Josh to find test programs, Steven to JUnit-ify them.

Example: https://github.com/unicode-org/unicodetools/blob/main/docs/idna.md
--> add an entry in the issue description/checklist above for TestIdna.java

Use judgment; some classes that should be unit tests may not be called *Test*.java.

Hi, just checking -- any progress? I think "Josh to find test programs" comes first...?

Thanks for the reminder -- I have been sidetracked but hoping to make some progress on this and #187 this week

@markusicu @srl295 I went through and added what appeared to be tests ("checks" in some cases).

Thanks @josh-hadley ! Over to @srl295 ?

• TestUnicodeInvariants should be shovel-ready, and it's important because it frequently shows us needing to adjust properties.
  • Ideal would be a parameterized test, I think, but even just initially having it fail (boolean) if there is anything inconsistent would be useful.
• TestIdna is hopefully easy
• CheckXmlProperties is definitely a test, not a tool.
  • As a test, it would be useful to JUnit-ify it.
  • Issue: The ucdxml files are usually updated only rarely, maybe only once or twice per release. This means that this test would usually fail, so we can't put it into Continuous Integration.
  • Issue: The docs suggest that the test is failing already, and that one is to ignore those failures. In order to make it pass, it sounds like we would need to improve the XML file parsing.
  • Issue: Ownership of the ucdxml files & spec is up for grabs. Future uncertain.
• CheckProperties looks more like a tool than a test
  • Most of its actions print useful or interesting information, or export properties data in another format (e.g., JSON), etc.
  • "COMPARE" shows the differences between the latest data and the previous Unicode version
  • "ICU" shows the differences between the current data and the ICU data
  • None of these actions look useful in JUnit form, and especially not for Continuous Integration
  • FYI: The docs show known systematic differences for Age values

@markusicu sounds good to me. I can do it along with #146 (similar work, separate PRs)

@srl295 do you have time to work on TestUnicodeInvariants & TestIdna ?

@srl295 do you have time to work on TestUnicodeInvariants & TestIdna ?

Generally yes maybe not today

I think #146 is done but need to verify

friendly ping @srl295 on TestUnicodeInvariants & TestIdna :-) (see #186 (comment))

It would be great to get these done soon, helping get the Unicode 15 beta ready.

@srl295 back in town?

@srl295 back in town?

I'll take a look tomorrow.

For a start, TestUnicodeInvariants does not have to become a parameterized test. The text above says "consider" :-) It would already be useful to just have it run as a test case and fail when either ParseErrorCount or TestFailureCount is not zero.

It looks like @srl295 turned TestIdna.java into a JUnit test already in PR #209, so I am checking that one off. (I then recently moved it from UnicodeJsps to plain unicodetools, and updated the instructions to say to run it as a JUnit test.)

For a start, TestUnicodeInvariants does not have to become a parameterized test. The text above says "consider" :-) It would already be useful to just have it run as a test case and fail when either ParseErrorCount or TestFailureCount is not zero.

I've scheduled it for my work tomorrow.

TestIdna.java

right, and it will also run along with all others when JUnit tests run.

@markusicu should CheckXmlProperties be JUnit-ified or not? The description and discussion seem to conflict

@markusicu should CheckXmlProperties be JUnit-ified or not? The description and discussion seem to conflict

Not. After this discussion we decided to skip it, and removed the check box from the description.

I will propose a small docs update before closing this issue.

🎉