jsonkeeper_migration.md

JSONkeeper database migration

old DB schema

CREATE TABLE "JSON_document" (
	id VARCHAR(64) NOT NULL, 
	access_token VARCHAR(255), 
	json_string TEXT, 
	created_at DATETIME DEFAULT CURRENT_TIMESTAMP, 
	updated_at DATETIME, 
	PRIMARY KEY (id)
);

new DB schema

CREATE TABLE "JSON_document" (
	id VARCHAR(255) NOT NULL, 
	access_token VARCHAR(255), 
	unlisted BOOLEAN, 
	is_json_ld BOOLEAN, 
	json_string TEXT, 
	created_at DATETIME DEFAULT CURRENT_TIMESTAMP, 
	updated_at DATETIME, 
	PRIMARY KEY (id), 
	CHECK (unlisted IN (0, 1)), 
	CHECK (is_json_ld IN (0, 1))
);

new columns:

• unlisted — could just be set to false because there will be no Activity Stream I guess
• is_json_ld — could just be set to false for the same reason

changes to existing columns

• length of ID column
• access_token — the new JSONkeeper requires a prefix in front of the access_token^[1]

[1] the old JSONkeeper just stores Firebase UIDs or X-Access-Tokens as is. To prevent collisions between to two types the new JSONkeeper adds prefixes frbs: and free: respectively. For exampe, instead of storing the Firebase UID gAo6iXt85qdWCZyCV3Dp8yxQH3R3, it would save it as frbs:gAo6iXt85qdWCZyCV3Dp8yxQH3R3 and instead of storing the freely chosen X-Access-Token foo would save it as free:foo.
I would assume that the X-Access-Token (since their is no client that uses it and therefore has to be given "manually" — e.g. when posting with curl) was only used by myself for tests. For a simple solution I would therefore suggest to just add the prefix frbs: to all access_token entries when migrating.