chris-braidwell

Converting DDL to pgroll format should be pretty simple using `pglast` in python: ```python >>> from pglast import parse_sql >>> stmt = parse_sql('alter table t1 ALTER COLUMN c5 TYPE TEXT;')[0].stmt...

> Error: `unable to execute start operation: pq: multiple primary keys for table "users_to_groups" are not allowed` This isn't a pgroll limitation. Postgres doesn't support multiple PKs in a table....

A good pattern, as long as the index is non-unique, goes like this: - outside of Prisma migration, run CREATE INDEX CONCURRENTLY ... - wait for that to finish -...

> The workaround I have allows me to use full automation from my repo, and that is: > > * use the `--create-only` option to let Prisma create a new...

I've found that a migration script that contains ONLY the `CREATE INDEX CONCURRENTLY` command will succeed. It appears Prisma doesn't wrap an additional transaction around a 1-command script. So having...