Upgrading PostgreSQL — reasons to

[brontolosone]

Upgrading PostgreSQL is not fun, but there may be pressing reasons to do it sooner than EOL of the current version. This issue is for accumulating such reasons. It may turn out that we won't accumulate sufficient reasons to upgrade before EOL, and that's good to know too!

Currently (September 2024) Central requires v14 (supported through November 12, 2026), v16 has been out for a year, v17 is probably not too far off.

Major reasons

• …

Minor reasons

• …

"Pity we didn't have this but we worked around it now already anyway"

• (pg 15) "Allow unique constraints and indexes to treat NULL values as not distinct" — would have allowed me to fold two very similar tables (user_site_preferences & user_project_preferences) into one, and would've saved some lines of code handling these two tables.

[alxndrsn]

For reference: https://www.postgresql.org/support/versioning/

[brontolosone]

Something to consider for migrating all Docker-users over on the next upgrade is to try to do it in-place.