jtyoung84

After perusing the [release notes](http://json-schema.org/draft/2019-09/release-notes.html) for draft 2019-09, `definitions` was renamed to `$defs` with the caveat that `definitions` should still be honored for backwards compatibility. This might be a non-issue...

It seems like an issue with the pydantic serializer. Even this causes issues: ```python from typing import Set from pydantic import BaseModel, ConfigDict from dataclasses import dataclass @dataclass(frozen=True) class Cell:...

Enums would be nice. I can't recall anymore exactly what the issue was that made me switch, but enums of objects were creating headaches somewhere. I do remember now that...

@bruno-f-cruz I'm going to assign this to @dbirman to see if this is a still an issue.

As an alternative, to make things backwards compatible, we can try something like: ```python from pydantic import BaseModel, Field, ConfigDict from enum import Enum from typing import Union, Optional class...

> Both of these solutions are inline with: > > > 3- Keep a service somewhere that takes a string and returns these complex objects. This is potentially very interesting...

``` { "$defs": { "Modality": { "enum": [ "behavior", "ecephys" ], "title": "Modality", "type": "string" }, "_Modality": { "additionalProperties": false, "properties": { "name": { "title": "Name", "type": "string" }, "abbreviation":...

> Some clarifying points about this PR: > > 1. This is separate from filename conventions. It is only meant to clarify how timestamps in the acquisition should be encoded....

> 1. Encode timestamps in local time w/ timezone offset (all necessary information in one place) This is **extremely** risky as this information can get lost in serialization/de-serialization cycles and...

> > Should we add a field for the location where the experiment was performed? That seems like a useful thing to track, and it would be simple to find...