I think that's an issue whatever service is sending you json right?

maybe we should make lenient readers read this?

@ZacSweers trailing commas are usually not allowed in Json, but since they are allowed in ECMA5 it would be really helpful if we have a feature of allowing it.

something along this Jackson trailing commas support

ECMA5 is 10 years old, is that still relevant today?

I agree that leniency could handle this if it's easy but it should remain invalid by default.

Lenient mode handles this. Only problem is we treat the comma as an absent null like org.json.. This is probably not what you want.

So lenient mode should handle trailing commas now (in v1.9.3) or by Icebox milestone? I'm trying to parse this JSON:

{
    "id": 1,
    "watermark": {
        "x": 772,
        "y": 1130,
        "width": 221,
        "height": 70,
    }
}

Moshi throws JsonEncodingException even in lenient mode. I tried various ways: calling lenient() on adapter, making a JsonReader instance with setLenient(true) ending up with the same result. The thing is I don't even need watermark object, I only care about some other fields (I cut them out for brevity, minimum reproducible example). Does it mean that lenient mode only allows commas in arrays, but not in objects?

Also, what is best practice workaround for this (considering there is no control over JSON source)? Modifying adapter, or sanitizing JSON before reading (regexp)?

@killvetrov we’re strict about these currently, even in lenient mode. We should fix this.

There is an PR since last year, why isn't it being reviewed? This issue is quite annoying