Re: run.language discrepancy between spec and schema

[James Kupsch <kupsch@cs.wisc.edu>]

I do not recall any discussion about this.  I agree that using ISO 639-1 
  and ISO 3166-1 for the value is sufficient.

Jim


On 4/27/19 2:18 PM, Michael Fanning wrote:
> Larry and I are (rather desperately) trying to button down the eballot 
> doc. We’ve found a difference between the spec and the schema that 
> neither of us can recall sufficient detail on to resolve. Anyone else 
> remember anything? Absent inputs, we’re favoring the use of ISO 639-1 
> and ISO 3166-1 for the reason that this data appears sufficient for 
> string mgmt. and it is a non-breaking change to expand to RFC5646 (but a 
> breaking change to move in the other direction).
> 
> Michael
> 
> The spec mandates RFC5646 to populate |run.language|. The schema 
> specifies that this property is constrained to a proper subset of 
> RFC5646, consisting of an ISO 639-1 country code and an optional ISO 
> 3166-1 region code.
> 
> Larry and I can't locate any history that helps us figure out how this 
> discrepancy materialized. The RFC5646 requirement is long-standing (in 
> fact, dates back to SARIF v1). The most likely scenarios are that we 
> either changed the requirement and failed to update the spec. Or that we 
> updated the schema speculatively in the SDK and someone made a 
> successful case against the change in TC discussion.
> 
> Does anyone recall the history on this? *@kupsch* 
> <https://github.com/kupsch>?
> 
> https://github.com/oasis-tcs/sarif-spec/issues/402
>