Two errors in the algorithm for calculating result.level

[ghost]

The last step in the algorithm in §3.27.10 for calculating result.level is:

IF level has not yet been set THEN
SET level to "warning"

This is incomplete for two reasons:

1. It doesn't say that you should only set level to "warning" if result.level is absent.

2. It doesn't say that if result.level is present, you should honor it, no matter what the rule metadata says.

[michaelcfanning]

Relates to #484 and a proposal should include fixes for both.

[michaelcfanning]

On second look at this, I think this could have been a candidate for the errata. The SDK issues cited in this tracking item relate to problems in the C# object model generated by that repository. i.e., there is no proposal/fix required in the standard itself - merely the clarifications that Larry suggested.

[schlaman-ms]

Document location for issue:

§3.27 result object
§3.27.10 level property

Last step in algorithm is incomplete:

lgolding:

"This is incomplete for two reasons:"

• It doesn't say that you should only set level to "warning" if result.level is absent.
• It doesn't say that if result.level is present, you should honor it, no matter what the rule metadata says.

Proposed clarifying last step in algorithm to resolve these two concerns.