This folder contains the database schemas for the RStudio Server database. These schemas are copied to the build output directory, and are applied to the database every time rserver is started.

Schema files rely on a file format convention containing the following:

• The date time at which the script was generated in the format YYYYMMDDHHmmssnnnnnnnnn (where nnnnnnnnn is nanoseconds)
• An underscore separating the date time from the friendly name
• A friendly name for what the script does
• A file extension, which can be .sql, .sqlite, or .postgresql. .sql files will be run for any database type, but the other two extensions are reserved for only running if the database in use corresponds to that file extension.

It can be cumbersome to generate the correct schema file names as they include a precise timestamp. For this purpose, simply run the make-schema.sh script from this folder, like so:

./make-schema AddNewTable sql

This will generate a file like 20200226141952248123456_AddNewTable.sql. To generate schemas files that are database specific, leave off the sql part of the command, which will generate a schema file for each supported database type.

You should not generate both a .sql and specific database extension types for the same migration. In this case, both migrations would be run, but this is nonsensical. You should either generate one .sql migration for the generic case, or one migration for each specific targeted database, not both.

In order to run any new schemas, you simply need to restart rstudio-server to ensure that any unapplied schemas are applied. Note that there is currently no way to rollback successful schema updates, so you are responsible for backing up your existing data if that is important to you.

Within your schema files, make sure to use snake_case naming for all table and column names. There are several reasons for preferring this. For more information, see https://github.com/rstudio/rstudio/issues/65899.