Re: [Codenarc-developer] Inline violations idea

[Chris M. <chr...@ea...>]

Thanks Artur. That is a clever and innovative approach for validation of
rule violations. Pretty cool! I acknowledge that configuring the violation
line numbers and source line text and keeping them in sync after edits can
be cumbersome. Your solution is a concise way to avoid that.

 

Some things to consider about the existing approach:

.         The existing approach and associated validations are there
intentionally to validate the exact line number in the generated Violation;
line numbers are not always completely straightforward and automatically
correct (especially around imports and annotations). Though the CodeNarc
framework and helper classes generally make that a non-issue now, I would
not want to skip line number validation for all rules/tests.

.         The source line text is not guaranteed to be the full source line
text; it may be a subset, or it may be null/empty. So, you could not
generalize that it is always the full source line.

.         I do like the way the existing approach allows a clean, standalone
example of source code that causes violation. It can also be easily copied
to the help documentation or pasted in an email (admittedly a minor
benefit).

 

That all being said, I do think your approach has value. I would want to
implement this as a separate method name, perhaps assertInlineViolations(),
rather than adding complexity (and potentially parsing the source code?) for
all invocations of assertViolations(). If you are willing to expand and
harden your POC as you mentioned, I think we can pull it in.

 

Thanks.

Chris