support for device custom property intervals

[brookfxg]

Changelog

Support for device custom property time intervals

Docs

Relevant docs will be updated with the associated API PR

Description

Pending

Adds support for device custom property time intervals - listing, getting, and updating. Also a small deduplication refactor to consolidate our device id or name handling into a single method. This is in line with how we do session id or key.

Testing

Local testing against a local API instance. A custom test script was used to hit the API with a high volume of reads and edits and examine the results for correctness.

[linear]

FG-14094 Update foxglove-python with device property history endpoints

[claude]

All prior inline threads resolved. The _device_identifier extraction is a nice dedup, and the urlquote addition to the existing device methods fixes path-encoding for names with special characters — that's a good side-effect improvement.

One note: the _device_identifier refactor changes the existing validation from truthy checks (if device_name and device_id) to is not None checks. This means callers who previously passed device_name="" alongside a device_id wouldn't have gotten the mutual-exclusivity error, but now will. More correct behavior, but technically a subtle breaking change for get_device, update_device, and delete_device.

PR description still has "Pending" in the Description section — clean that up before un-drafting.

[claude]

suggestion: The list test covers key, limit, offset, and projectId but doesn't exercise start/end. Those params go through astimezone().isoformat() serialization, which is the most interesting codepath to verify (timezone handling, format). Worth adding a test that passes start/end datetimes and asserts the query params are serialized correctly.

[claude]

Prior inline threads all resolved except the start/end test coverage suggestion — still worth adding.

Two new questions on the update method (see inline): projectId placement (body vs query param) breaks from the pattern every other device endpoint uses, and the 204/else branching looks like it might have a dead code path.

mguida22's suggestion to combine _device_identifier + urlquote into a single helper is still open — Brook's proposed _device_identifier_with_path sounds reasonable, worth doing before merge to avoid the repetition across 6 call sites.

PR description still has "Pending" in the Description section.