In describing standard capabilities, the specification makes no distinction between capabilities provided by the local end and those provided by the remote end. This is generally fine because they largely represent the same concept in both contexts.
However, the "browserVersion" capability differs--the value provided by the local end may describe a range of versions, while the value provided by the remote end describes exactly one version.
This distinction is only apparent in the "matching capabilities" algorithm. While I doubt any protocol-level change is necessary here, I'm curious if the editors have any opinions about if/how this distinction can be communicated more clearly. For instance, the current description of the capability is inaccurate:
In describing standard capabilities, the specification makes no distinction between capabilities provided by the local end and those provided by the remote end. This is generally fine because they largely represent the same concept in both contexts.
However, the "browserVersion" capability differs--the value provided by the local end may describe a range of versions, while the value provided by the remote end describes exactly one version.
This distinction is only apparent in the "matching capabilities" algorithm. While I doubt any protocol-level change is necessary here, I'm curious if the editors have any opinions about if/how this distinction can be communicated more clearly. For instance, the current description of the capability is inaccurate: