feat(PostgreSQL DB Storage): Added support for Postgres. Changes incl… #635
Conversation
…ude liquibase changesets to target postgres with test running against both MySQL & Postgres in test containers
| databaseChangeLog: | ||
| - changeSet: | ||
| preConditions: | ||
| onFail: CONTINUE | ||
| dbms: | ||
| type: postgresql | ||
| id: create-indexes-postgresql |
There was a problem hiding this comment.
This changeset is similar to the MySQL one except that PgSQL indexes are globally scoped and we can't have same index name for every table and this changeset is applying that rule.
There was a problem hiding this comment.
What happens with the schemas defined by MySQL with the duplicate index names? Does pg still create the indexes but they're wrong, or what? I would expect that if the indexes conflict, it would generate an error and fail to apply the changeset, so we'd never land here?
There was a problem hiding this comment.
PG will fail to execute the changeset.
There was a problem hiding this comment.
Not sure what you meant by so we'd never land here , but our original changeset that we had will fail for postgres and that's the main reason I had to add preconditions in there.
front50-sql/src/main/resources/db/changelog/20190530-add-is-deleted-indexes.yml
Show resolved
Hide resolved
| @@ -0,0 +1,89 @@ | |||
| databaseChangeLog: | |||
There was a problem hiding this comment.
This is a new changeset only applied for Postgres. CHAR types are converted to bpchar (blank pad characters) in Postgres and they are not recommended datatypes + JOOQ doesn't play well with padded column values. So this change set modifies the column data type to VARCHAR.
There was a problem hiding this comment.
I did a quick look around but came up short - could you link where it's documented that char / bpchar is not recommended? I'd like to get more learned on this.
There was a problem hiding this comment.
dbmstype./cc @cfieber @robzienert @ajordens