Re: [calsify] WGLC JSCalendar

[Michael Douglass <[e-mail]>]

On 10/27/19 20:28, Neil Jenkins wrote:
> On Sat, 26 Oct 2019, at 03:01, Michael Douglass wrote:
>> What I meant was that the objects in the list should have a type:
>>       "localizations": {
>>         "de": {
>>           "@type": "PatchObject"                    <------------------- this
>>           "title": "Live von der Music Bowl: The Band!",
>>           "description": "Schau dir das groesste Musikereignis an!",
>>           "virtualLocations/6f3696c6-1e07-47d0-9ce1-f50014b0041a/name":
>>                                   "Gratis Live-Stream aus der Music Bowl"
>>         }
>>       }
>>
>> I think patch objects are the only objects that don't have a type. I 
>> think for simplicity and ease of validation all objects should be 
>> required to have @type
>>
>
> No. The |@type| property is for objects with a named set of attributes 
> (essentially a class). The PatchObject is a primitive datatype, more 
> like Integer etc. There are no predefined keys; each key is a patch to 
> apply. So if you add |@type| there, that's actually a path to 
> overwrite the |@type| property on the object you apply it to!
>
I still don't agree.

There's no way this is a primitive type. It's an object whichever way 
you look at it. The fact there are no predefined keys is an attribute of 
the pathObject class. Without the @type I have to defer any validation 
until the whole structure is parsed - or - for this one case only - I 
have to pass down the expected type.

As for the @path - In section 4.3.3 (version 21) for recurrence 
overrrides we have

    A pointer in the PatchObject MUST be ignored if it starts with one of
    the following prefixes:

    o  @type
and a bunch more

I think we could make that ignoring @path a general rule for pathObjects 
- is there anywhere we want to allow changing the type?