Paul "Code Grump" Turner

Regarding #31: I'm bumping up against the limitations of the Gherkin parser, most notably a lack of precise token positions used for source-mapping. It would be really good to have...

> AFAIK ValueTask is also supported. I based this off the constraints in our documentation. If `ValueTask` is supported, the documentation should probably be updated to reflect that. In principal...

Unfortunately, that's not reflected in the documentation regarding step definitions: https://docs.reqnroll.net/latest/automation/step-definitions.html > ### Step Definition Methods Rules > - Must be in a public class, marked with the `[Binding]` attribute....

How would you feel about your test methods accepting a CancellationToken passed in from the NUnit runner to signal the test execution should be cancelled?

It will be sufficient, provided the steps are only used within a single feature. If that step is used by another feature, it will only prove that we can run...

I've created this PR at this stage for feedback on what exists so far. I would suggest the following types should be given the most attention: - `SyntaxNode` and `RawSyntaxNode`...

> To get a design overview of the change, do I see it correctly, that this change basically uses the Gherkin parser, but provides an alternative AST builder and token...

I've dropped in a single example into the associated test project which demonstrates a hypothetical analysis: detecting a lack of a blank line between the feature headline and its associated...

Regarding incremental parsing support: there's two things the design of this model does to support this future desire. Firstly, the raw nodes do not have any concept of "ownership"; they...

> > Regarding incremental parsing support: there's two things the design of this model does to support this future desire. > > Maybe it's not relevant, but how is this...