wiki: invalid sample for NUnit 3.0

[ghost]

https://github.com/nunit/docs/wiki/XML-Formats -> http://nunit.org/files/testresult_30.txt

e.g. of error: "environment" must be child of "test-suite", but in the sample its parent is "test-run"

There might be more errors, maybe add a more recent sample...

[jnm2]

@nunit/framework-team Would the best approach be to compare current framework code to the docs, or maybe to even come up with an XSD in the framework repo?

[mikkelbu]

I say both. We already have an issue here and in the framework repo about the schema AFAICR

[jnm2]

@mikkelbu I have some extra time today. Was anyone else hoping to tackle it?

[ChrisMaddock]

@jnm2 - The XSD issue has been sitting as something we think we should do but nobody wants to since 2014. I reckon it's all yours! ;-)

https://github.com/nunit/nunit-console/issues/13

[jnm2]

😆 Okay. Now I'm confused. I would have thought that the XML was driven totally by the framework, not the engine. Where am I looking to find the source of the XML generation then? In other words, which repo will the xsd reside in?

[ChrisMaddock]

The XML per-assembly is created in the framework. The engine merges these multiple xml results into a single file (e.g. wraps them in a test-run element.)

I would imagine we'd want an xsd in both repos! I'm not sure if it's possible to support 'nested XSDs'?

[jnm2]

Okay! I'll put each thing as close to the defining code as possible. I believe you can include and import XSD, but I'd have to read up on it.

[ChrisMaddock]

Sounds awesome! 😄

[jnm2]

Is there a target namespace we have used in the past?

I'm also leaning towards xs:all everywhere rather than xs:sequence, which gives us the freedom to change the order in which elements appear. Parsers would need to use a string lookup either way because some elements will have minOccurs="0". What do you think?

(I'm writing this with descriptiveness, not prescriptiveness, in mind.)

[ChrisMaddock]

I'd agree - the structure should be important, the order, not so.

I don't know about namespaces. I think NUnit 2 used to have an XSD - might be worth taking a look over at nunit-legacy.

[jnm2]

Here's what I got so far, validates against real-world test XML from the Explore action:

https://github.com/jnm2/nunit/blob/xsd/src/NUnitFramework/framework/Schemas/Test.xsd#L1

That has to come first, because the result format extends this format. What do you think so far?

[ChrisMaddock]

I'm not very familar with xsd myself, but looks good to start with! This doesn't cover any failure-cases right, yet? e.g. throwing an exception during explore?

[jnm2]

Oh, good question! I hadn't thought of that.

[jnm2]

@ChrisMaddock Any exceptions thrown during explore are already covered by the normal result format. 👍