The report having a property named report is confusing

[RByers]

This is a silly thing but it's been bugging me. Mostly the spec uses the term report to refer to the larger thing (eg. that has a type) and body to refer to the the type-specific data structure. But then the exposed property name for the body is "report".

This leads to awkward observer code like report.report.

Can we just rename the property to body to match the spec terminology?

Alternately, since we likely want a WebIDL definition for each report type (at least those exposed via #29), we could recast the spec in terms of those types instead of a generic notion of "any objection which can be serialized to JSON" (which itself is problematic), and then the body could just be contained directly in the report via interface inheritance:

interface ReportBase
{
    readonly attribute string type;
};

interface DeprecationReport : ReportBase
{
    readonly attribute string message;
};

@mikewest @juliatuttle

[mikewest]

Renaming the property to body sounds perfectly fine to me. I think I like that clear separation between the generic reporting API stuff, and the stuff specific to a specific thing that's being reported. It also resolves the question that will inevitably crop up when a feature has a property that collides with a generic reporting API property (e.g. assume your example had deprecation "types").

[RByers]

Thanks SGTM. PTAL at #43

[RByers]

By the way, I think we might still want to define the reports in terms of WebIDL interfaces instead of just "JSON serializable ECMAScript objects", but we can discuss that separately as part of the spec work for #29.