Manage ids across different mappings connected to the same database

[spine-o-bot]

In GitLab by @manuelma on Sep 29, 2018, 15:26

There needs to be a way for a DiffDatabaseMapping to reserve ids, so that other processes don't steal them from it. An idea would be to create a table next_id that the DiffDatabaseMapping will keep updated. If the other process finds this table, it'll also use it and update it.

[spine-o-bot]

In GitLab by @manuelma on Sep 30, 2018, 15:19

assigned to @manuelma

[spine-o-bot]

In GitLab by @manuelma on Sep 30, 2018, 19:12

mentioned in commit 75a395e

[spine-o-bot]

In GitLab by @manuelma on Oct 1, 2018, 10:45

The next_id table is in place and works fine. Needs testing. One weakness is ids are incremented before the user commits their changes, so if they rollback instead we 'lose' a lot of ids. Do we care?

[spine-o-bot]

In GitLab by @jkiviluo on Oct 2, 2018, 10:30

You mean there will be id numbers that have been skipped over and will not be used in the DB anymore? I wouldn't care as long as the max id is substantial.

[spine-o-bot]

In GitLab by @manuelma on Oct 2, 2018, 10:48

Apparently the max in SQLite is 2^64, that is 1.84e19. So we have some room but also we can implement some logic so as not to skip ids.

[spine-o-bot]

In GitLab by @jkiviluo on Oct 2, 2018, 14:54

In my opinion that's substantial. :)

[spine-o-bot]

In GitLab by @manuelma on Jan 24, 2019, 16:07

This is done but needs unittesting.