Refactor projects URLs to use ID or UUID as identifier rather than project name

[dboehmer]

When I designed the URL scheme for projects like /project/my-vacation for project My Vacation I used GitHub as role model. This doesn't seem to fit:

• Coocook users need no "speaking" URLs with the literal project name like the Git URLs used for syncing with GitHub
• a Coocook project has no immutable relation to 1 user like a GitHub repo belongs to a GitHub user—so we already have no username in the URL
• so we have 1 common namespace for projects which sucks

Instead we should mimic YouTube URLs like https://youtu.be/DLzxrzFCyOs:

• URLs just contain 1 identifier
• identifier might be database ID or UUID
• identifier is encoded with base64 or similar

Then URL should look like e.g. /project/A1b2z3/settings.

Pros:

• users can have projects of same name
• 1 user can have multiple projects of same name—should this be permitted??
• Bonus: projects can't can be renamed without problems.

Stackexchange Webapps: format of the YouTube video ID

[dboehmer]

note:

YouTube video ID is a 8 byte/2⁶⁴ integer encoded as base64 without padding characters.

The YouTube channel/playlist IDs are 16 byte/2¹²⁸ integers.

Standard SQL supports the BIGINT datatype which is 4 bytes 8 bytes large and allows values up to 2⁶⁴ = 18,446,744,073,709,551,615 (unsigned). This is supported both by SQLite, PostgreSQL and MariaDB. See https://en.wikipedia.org/wiki/Integer_(computer_science)

• https://sqlite.org/datatype3.html
• https://www.postgresql.org/docs/current/datatype-numeric.html
• https://mariadb.com/kb/en/bigint/

[dboehmer]

I just merged a branch into master that realized phase I of this migration. For now we use the database IDs in decimal format. I just couldn't decide for any other system. A sequence of decimal IDs is still much better than the user-provided natural strings that share the same namespace.

In a next step we can either

• encode a decimal ID, e.g. as base32, possibly encrypted with a secret seed for secrecy, or
• add another column with a random ID

I am confident that we can manage to migrate from phase I to phase II smoothly.