Allow json5 routines and configurations

[thclark]

Feature request

It would be useful to be able to comment the configuration files, is there any easy way of accepting JSON5, with its inbuilt comments?

Example comments in a routine.json file:

{
  "commands": [
    ["startSync", 0],
    ["selMsgDest 3", 0.1],
    ["getBattery", 0.2],
    ["startBaros", 0.3],
    ["selMsgDest 4", 1.1],
    ["getBattery", 1.2],
    ["startBaros", 1.3],
    ["selMsgDest 5", 2.1],
    ["getBattery", 2.2],
    ["startBaros", 2.3],
    ["selMsgDest 3", 60.1],
    ["stopBaros", 60.2],
    ["selMsgDest 4", 61.1],
    ["stopBaros", 61.2],
    ["selMsgDest 5", 62.1],
    ["stopBaros", 62.2],
    ["stopSync", 65],
    ["enterHighSpeed", 65.1],
    ["selDevice 3", 70], // (wait until the connection parameter is updated, you can tell that with the constatpacket number)
    ["readSens", 75], // allow 1 minute download per minute collection
    ["selDevice 4", 135], // (wait until the connection parameter is updated, you can tell that with the constatpacket number)
    ["readSens", 140],
    ["selDevice 5", 200], // (wait until the connection parameter is updated, you can tell that with the constatpacket number)
    ["readSens", 205],
    ["exitHighSpeed", 265],
    ["selMsgDest 3", 270.0], // (now we reconnect all the nodes and put them into sleep)
    ["sleep", 270.1],
    ["selMsgDest 4", 271.0],
    ["sleep", 271.1],
    ["selMsgDest 5", 272.0],
    ["sleep", 272.1],
    ["selMsgDest 3", 330.0], // (wake up nodes and allow them to re-establish connection parameters)
    ["wakeUp", 330.1],
    ["selMsgDest 4", 331.0],
    ["wakeUp", 331.1],
    ["selMsgDest 5", 332.0],
    ["wakeUp", 332.1]
  ],
  "period": 360
}

[cortadocodes]

Would this require us using pyjson5? It's apparently is up to 6000 times slower than pyjson.

Alternatively, we could write something (or use an existing solution) that will strip the comments before they're parsed by pyjson.

Or, if this is just for examples, we could just include it in the docs in a code-block.

[Kijewski]

I maintain pyjson5 (which is different from json5), and has the same or a better performance than the builtin json in many cases: https://pyjson5.readthedocs.io/en/latest/performance.html

[cortadocodes]

Ah my mistake! Are those two links supposed to go to the same page? I drew my conclusion from the description on this page

[Kijewski]

Oops, I copied the wrong link. https://pypi.org/project/pyjson5/ is written in cython, so it's quite fast. https://pypi.org/project/json5/ is written in pure python, so it's quite slow.

[cortadocodes]

Cool cool! I think we can look at using pyjson5 then