Add nl.basjes/yauaa_context/jsonschema/2-0-0

[oguzhanunlu]

This PR upgrades yauaa_context schema to 2-0-0 by converting enums to strings with a future-proof maxLength(128). We don’t have control on those enums and any new version of yauaa lib can add and remove items.

Why bump to 2-0-0?
coming soon

$ diff schemas/nl.basjes/yauaa_context/jsonschema/1-0-4 schemas/nl.basjes/yauaa_context/jsonschema/2-0-0
8c8
<         "version": "1-0-4"
---
>         "version": "2-0-0"
14c14,15
<             "enum": ["Desktop", "Anonymized", "Unknown", "UNKNOWN", "Mobile", "Tablet", "Phone", "Watch", "Virtual Reality", "eReader", "Set-top box", "TV", "Game Console", "Home Appliance", "Handheld Game Console", "Voice", "Car", "Robot", "Robot Mobile", "Spy", "Hacker", "Augmented Reality", "Robot Imitator"]
---
>             "type": "string",
>             "maxLength": 128
45c46
<             "enum": ["Desktop", "Mobile", "Cloud", "Embedded", "Game Console", "Hacker", "Anonymized", "Unknown"]
---
>             "maxLength": 128
67c68
<             "enum": ["Browser", "Mobile App", "Hacker", "Robot", "Unknown", "Special", "Cloud", "eReader"]
---
>             "maxLength": 128
96c97
<             "enum": ["Browser", "Browser Webview", "Mobile App", "Robot", "Robot Mobile", "Cloud Application", "Email Client", "Voice", "Special", "Testclient", "Hacker", "Unknown", "Desktop App", "eReader"]
---
>             "maxLength": 128
139c140
<             "enum": ["Weak security", "Strong security", "Unknown", "Hacker", "No security"]
---
>             "maxLength": 128
153a155,158
>         "webviewAppNameVersion": {
>             "type": "string",
>             "maxLength": 1000
>         },

ref: PDP-1090

[istreeter]

My only doubt is whether 32 is long enough to future-proof the new fields.

In #1216 we got bored of forever bumping max lengths, whenever the yauaa library changed, or whenever a new user agent got invented. So we deliberately bumped all max lengths to 128 or more, to better future-proof the schema.

[istreeter]

Looks good from my point of view.

@oguzhanunlu please can you remind me why this is 2-0-0, not 1-0-5? I'm sure you're probably right, but I've forgotten the reason.

Even better... you could add the explanation to the PR description, so we have a long term reference of why it is 2-0-0.

How urgently does this need to be merged? Presumably you can you use the dev registry while working on the feature in Enrich?

[oguzhanunlu]

Thanks @istreeter ! I just updated PR description with a short note on why it is 2-0-0 from my point of view. There is no urgency as I can reference my branch when needed. Let me add that this is a blocker for the next enrich release which is coming in the next few weeks.

[istreeter]

payloads of new schema will always be invalid against 1-0-4, hence a model bump.

This is not the criteria for a model bump. We often add new extra fields to iglu central schemas, and we do it as an addition (e.g. 1-0-0 --> 1-0-1). The original schemaver blog post tells us so.

Now for this PR.... you might be correct it should be a 2-0-0 schema, but it's not because you added the webviewAppNameVersion field. It's because you changed several fields from enum to regular string. If there is any chance that might break any loader, then this should be a 2-0-0 schema. That is the point I want to see documented in this PR description.

[oguzhanunlu]

thanks @istreeter ! my bad, I was asking the wrong question. The blog post asks Will this new schema validate all historical data?. Here all historical data will be valid against new schema as the new field isn't required and additionalProperties is false, so it indicates an addition on its own.

It's because you changed several fields from enum to regular string. If there is any chance that might break any loader, then this should be a 2-0-0 schema.

Asking the same question. Since previous enum possibilites are always string with length <128, it looks like new schema can validate all historical data. So it indicates an addition as well. However I'm not sure if type conversion could break some of the loaders, I need to check them.