5.5.3 Encodings
Typo in para 1, should "This is a list of strs. The val of each str"
9.4 Metadata
This is a better design than the last version. But it still seems unsatisfactory when it comes to how it will be applied to Haystack tags. In Haystack the vast majority of tags are marker tags which the proposed oBIX spec requires those tags to be sub-objects. But in reality marker tags are much closer to oBIX contracts. I think it would be cleaner to define marker tags in the "is" attribute using their tagspace as their prefix. Consider:
// proposed version
<obj displayName="Discharge Temp Sensor">
<obj ts="h" name="discharge"/>
<obj ts="h" name="temp"/>
<obj ts="h" name="sensor"/>
<obj ts="h" name="point"/>
</obj>
// versus simpler design
<obj displayName="Discharge Temp Sensor"
is="h:discharge h:temp h:air h:sensor h:point">
</obj>
14.1 History Object - HistoryCollection
The terms "collection" and "HistoryCollection" seem sort of generic and confusing. How about "collectMode" and "HistoryCollectMode"
14.1 History Object - prototype
Glad to finally see something like this. However I am not quite sure how it works. I think we need an example.
14.2.4.2 History Query as Preformatted List
This design requires two separate round trip requests which is pretty terrible for performance. I think a requirement of this design is that once the history point is learned by the client system, that it can request history data with a single HTTP request. I don't see that having an oBIX XML response before getting the data provides much value. I would suggest this: just add a "format" field to the HistoryFilter object and when specified the result is returned in the requested MIME type.