Re: let ALTER COLUMN SET DATA TYPE cope with trigger dependency - Mailing list pgsql-hackers

From jian he
Subject Re: let ALTER COLUMN SET DATA TYPE cope with trigger dependency
Date
Msg-id CACJufxFjSGY2npNQjqsFxqa7H5tN-rRV1NuxXmFicGm+N-kh-w@mail.gmail.com
Whole thread Raw
In response to let ALTER COLUMN SET DATA TYPE cope with trigger dependency  (jian he <jian.universality@gmail.com>)
List pgsql-hackers
hi.

v2-0001, v2-0002 is the same as [1].
The parameters relOid and refRelOid are removed from the CreateTrigger function
and instead added to the CreateTrigStmt structure.

The main reason is to avoid repeated name lookups.
In fact, we already did this in CreateTriggerFiringOn, see below code snippet.
If the OID is valid, use it preferentially; otherwise, fall back to resolving
the relation via the RangeVar.

```
    if (OidIsValid(relOid))
        rel = table_open(relOid, ShareRowExclusiveLock);
    else
        rel = table_openrv(stmt->relation, ShareRowExclusiveLock);

        if (OidIsValid(refRelOid))
        {
            LockRelationOid(refRelOid, AccessShareLock);
            constrrelid = refRelOid;
        }
        else if (stmt->constrrel != NULL)
            constrrelid = RangeVarGetRelid(stmt->constrrel, AccessShareLock,
                                           false);
```

v2-0003: is the actual implementation.
The general workflow would be as follows: collect all affected triggers,
retrieve their definitions, drop the existing triggers, and then recreate them.
Any comments associated with the triggers will also be dropped and recreated.

[1]: https://postgr.es/m/CACJufxFu7Y4FhVkaKT2Kaj8ym2T5TcwN93cR_6h4x66iLrSZ-Q@mail.gmail.com


--
jian
https://www.enterprisedb.com/

Attachment

pgsql-hackers by date:

Previous
From: jian he
Date:
Subject: Re: CREATE TABLE LIKE INCLUDING TRIGGERS
Next
From: "zengman"
Date:
Subject: Re: [Patch] timezone/zic.c: Fix file handle leak in dolink()