[core] SenML JSON syntax (with multiple base objects)

[Ari Keränen <[e-mail]>]

Hi CoRE WG folks,

Final SenML open issue for today is whether we should have simpler format for the JSON. Currently the draft uses nested array for the measurement entries as discussed in the Prague meeting (see for example [1]). However, it could be simpler to use just a single array and have entry and base objects at the same level. 

For example:

[
 {"bn": "urn:dev:mac:0024befffe804ff1/current",
 "bt": 1276020076,
 "bu": "A"},
  {  "t": -3, "v": 0.14e1 },
  {  "t": -2, "v": 1.5 },
  {  "t": -1, "v": 1.6 },
  {  "t": 0,  "v": 1.7 },
 {"bn": "urn:dev:mac:0024befffe804ff1/voltage", "bu":"V" },
  {  "t": -2, "v": 120.05 },
  {  "t": -1, "v": 120.1 },
]

The base objects would use the same JSON merge patch format and logic as before, but all entry objects would be simply inside the root array, potentially intermixed with base objects.

One challenge with this format is that when parsing, one does not immediately know whether it is a base object or entry object, but this should be possible to discover from the member names ("b*"). Of course to accommodate for new base and entry members, we would need to have logic that makes the difference clear even if there are unknown members in the objects (for example, "only b* member names allowed in base object and no b* in entries").

Seems sensible to me. What do you think?


Cheers,
Ari

[1] https://tools.ietf.org/html/draft-jennings-core-senml-01#section-6.1.2