Feature/on delete

[antoineB]

No description provided.

[matthias-Q]

Thanks for the PR. I made some remarks that need to be addressed.

[matthias-Q]

this is already part of #196 right?

[antoineB]

Yes I used my previous MR to avoid conflicts.

[matthias-Q]

This is already part of #196

[antoineB]

Yes.

[matthias-Q]

This is already part of #196

[matthias-Q]

this could be combined in one optional block with a choice between keyword_delete and keyword_update. I see no difference between the other parts of the sequence.

[matthias-Q]

Thanks for the changes. I double checked the statement in the Postgres documentation and it looks like a small part is missing. Would be nice if you could add this.

[matthias-Q]

I think set null and set default (this one is missing), can optionally be followed by a column name. In our parser the column name is represented by the object_reference node. (paren_list() of object_reference

https://www.postgresql.org/docs/current/sql-altertable.html (under referential_action)

[antoineB]

You cannot put an object_reference just an identifier:

toto=# create table test1(id integer references test0(id) on delete set null(test.id));
ERROR:  syntax error at or near "."
LINE 1: ...d integer references test0(id) on delete set null(test.id));

[antoineB]

Also line 2273 $.ordered_columns should be paren_list($.identifier) at least on postgres.

[matthias-Q]

This worked for me on postgres 15.3

CREATE TABLE table1 (
    id serial PRIMARY KEY,
    column_name1 integer,
    column_name2 integer
);

CREATE TABLE table2 (
    id serial PRIMARY KEY,
    table1_id integer REFERENCES table1(id) ON DELETE CASCADE
);

ALTER TABLE table1
DROP CONSTRAINT IF EXISTS table1_column_name1_fkey;

ALTER TABLE table1
ADD CONSTRAINT table1_column_name1_fkey
FOREIGN KEY (column_name1)
REFERENCES table2(id)
ON DELETE SET NULL (column_name1);

[matthias-Q]

You are right about object_reference. This won't work. identifier does work however.

[matthias-Q]

@antoineB Thanks for the MR. I have two general remarks:
Would be nice if you would not force push changes that are requested in a review. It makes it easier to review. We can always squash commits before merging. It would also be nice if you could add a small query that is failing or not yet supported and also provide the link to the SQL Dialect documentation. This makes it a lot easier for us.

However, we should definitely think about some contribution rules