Fix AUTOINCREMENT

[MartinRied]

AUTOINCREMENT was not working with the previous implementation, since the actual AUTOINCREMENT/IDENTITY keyword was missing.

Based on other examples, this should usually be provided by AbstractJdbcDatabase#getAutoIncrementClause().

On the other hand, the SnowflakeDatabase#getAutoIncrementClause(BigInteger,BigInteger) method is as far as I can tell irrelevant, since it's not exposed or used by AbstractJdbcDatabase or the Database interface.

The updated auto-increment code has been tested against a live Snowflake instance to create tables with auto-increment enabled.

┆Issue is synchronized with this Jira Bug by Unito

[MartinRied]

Hi @molivasdat, @nvoxland,

Is there anything I could do to support the review process?

We're using liquibase-snowflake internally and so far I've avoided building our own version in favor of waiting for the PR to hopefully be approved - but of course this blocks any CI/CD which depends on the modified liquibase-snowflake version.

Regards,
Martin

[molivasdat]

Hi @TheDECKER Do you have an example changeset that can be validated against this change? And thank you for adding to the automated tests.

[MartinRied]

Hi @molivasdat ,

I've slightly modified one of our tables accordingly - though I have to admit I didn't re-test that changeset against Snowflake using the "vanilla" liquibase-snowflake code.

https://pastebin.com/vF5cuevR

[molivasdat]

Thanks for the update. I noticed that you have constraint primary key which by definition should be unique but also have unique=false on line 11.

[MartinRied]

It certainly looks a bit odd when you read it :-D

On the other hand, what I know from Oracle & co is that when you create a PK constraint, it is automatically unique, so generating a separate UNIQUE constraint should not be necessary. I've got to admit, though, that I'm not sure what ANSI SQL has to say about that...

And lastly:
For Snowflake it really doesn't matter at all, since it allows creating PK and UNIQUE constraints but does not enforce them!

[MartinRied]

Actually I came across the fact that even when using AUTOINCREMENT & UNIQUE & PK on a Snowflake column I can:

• Manually INSERT a record with PK=1
• Manually INSERT another record with PK=1
• INSERT a record without PK - triggering the AUTOINCREMENT - which will produce a third record with PK=1 when I do it for the first time.

All without raising any constraint exceptions..