Change the definition of device.id

[bencehornak]

Fixes #1874

Warning

This PR must be merged together with #1897 as they are highly related, dear maintainers, please add them to the same milestone.

Changes

I have given a new definition to device.id, which now truly identifies the device. It has been marked as opt-in though, as its usage is generally discouraged due to privacy implications, and should only be enabled in special cases, when it does not violate the privacy of the user.

The old definition of device.id is more-or-less covered by the proposed app.installation.id in #1897, which is recommended for general use instead of this, as it provides more privacy to the users.

Merge requirement checklist

• CONTRIBUTING.md guidelines followed.
• Change log entry added, according to the guidelines in When to add a changelog entry.
  • If your PR does not need a change log, start the PR title with [chore]
• schema-next.yaml updated with changes to existing conventions.

[bencehornak]

The markdown-link-check step fails at this step, because this file will only be available after #1897 is merged.

[trask]

cc @open-telemetry/semconv-mobile-approvers

[trask]

cc @open-telemetry/semconv-mobile-approvers

hm, it looks like CODEOWNERS is set up so that @open-telemetry/semconv-mobile-approvers should have been added as reviewers on this PR 🤔

UPDATE oh, my bad, they were: #1951 (comment)

I guess I didn't realize (forgot?) that the group name is replaced once someone from that group reviews it

[trask]

it would be great if we can get approvals from the iOS and browser perspectives as well (if this attribute is applicable there)

[MSNev]

As trask requested feedback via the Client Sig this morning, for current SDK's we support the sending / population of this either as

• By default for a browser as the word "browser", so as with the "example" value of "123456789012345". "browser" is another "possible" value.
• Other internal SDK's allow the setting of this value based on any supporting hardware lookup (like when running from react-native), in which case it matches the description as specified.

[bencehornak]

@MSNev do you think the browser value is desired? To me it doesn't really sound like an 'identifier', as it's not trying to provide uniqueness.

[MSNev]

Suggestion: As a generic "namespace" for devices, its probably also worth considering an additional device.class which describes the type of device in a more generic form with possible values like

• Mobile
• Desktop
• Android
• Windows
• XBox
• iOS
• Browser
• etc

Although, there may be a collection of other attributes which could be used by backends to derive some of this information.

[bencehornak]

Sounds reasonable to me, I would keep this suggestion to a separate PR though, if you don't mind.

[bidetofevil]

The new description sufficiently neuters device.id and gives the appropriate warnings for commercial apps and SDKs to NOT user it, so this works for me.