michaelgira23 / crumbsdb Goto Github PK

Currently, the plan is to have the user initialize all of the tables they want to use (including table name and column names and types). If the table matches an existing table stored in the cookie, it will simply use the data from the existing table. If it does not match any existing table, it will create a new one.

The problem is what happens if a table has the same name as an existing table, but different column names or types. Here are a few options:

1. Try to handle existing table as best as possible: any existing columns will be translated over to the new table, and existing columns not in the new configuration will be dropped. Any new columns will use default values. If no default values provided, entire row will be dropped completely.

2. Delete existing table as if nothing ever happened

Most likely go with option 1.

This same issue is present for the crumbsDB version number inserted at the beginning of the cookie (Ex. [crumbsDB 1.0.0]).

1. Try to handle as gracefully as possible: Try to make most sense out of the data if possible. Similar to option 1 above. If can't parse database, just drop it.

2. Delete old table

Most likely, we will go with option 2 if the major version is different (Ex. 2.0.0 vs 1.0.0) following the semver convention.

We must also decide whether to keep old crumbsDB data. Should we store the cookie as {TABLE_NAME}-{CRUMBSDB_VERSION} in case crumbsDB is reverted back?