RE: test1149 and test1249 [was RE: New test]

Glen,

> I'm not exactly sure why this should be any different than, for
> instance, tests 1140 and 1143... 1140 tests a required feature, and 1143
> tests the same required feature but for a different case.  They're both
> important and both marked REQUIRED.

yup, a couple of weeks ago I'd have made it REQUIRED.

> Just because the assertions are already "covered" by different tests
> doesn't mean that this test isn't checking a piece of required behavior,
> and in fact one which most implementations are currently doing
> incorrectly.

agreed, 100%!

> Personally, I'd leave it REQUIRED, since it's not testing an optional
> feature and it covers a case the other tests don't.  *shrug*

In terms of the report, cr='REQUIRED' means we want to demonstrate the 
testcase interoperating between 4 implementations before moving the 
Core and SOAP specifications from CR to PR, 'OPTIONAL' means 2 
implementations should pass and 'INFORMATIONAL' is a bucket for all 
other testcases which have merit.

If we have another round of testing for CR, then I'd agree 
making this 'REQUIRED' adds good value, as it is I think it would 
invalidate the CR report for little gain in terms of the spec. 

Paul

Received on Monday, 20 March 2006 17:12:23 UTC