Horton

Re: cataloguing NOT NULL constraints

Peter Eisentraut <peter.eisentraut@enterprisedb.com>

View thread
From: Peter Eisentraut <peter.eisentraut@enterprisedb.com>
To: Alvaro Herrera <alvherre@alvh.no-ip.org>, Pg Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2023-03-03T10:15:00Z
Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Revert structural changes to not-null constraints

  2. Fix inconsistencies in error messages

  3. Disallow direct change of NO INHERIT of not-null constraints

  4. Disallow NO INHERIT not-null constraints on partitioned tables

  5. Better handle indirect constraint drops

  6. Don't try to assign smart names to constraints

  7. Fix restore of not-null constraints with inheritance

  8. ATTACH PARTITION: Don't match a PK with a UNIQUE constraint

  9. Fix propagating attnotnull in multiple inheritance

  10. Check stack depth in new recursive functions

  11. Move privilege check to the right place

  12. Update information_schema definition for not-null constraints

  13. Fix not-null constraint test

  14. Disallow changing NO INHERIT status of a not-null constraint

  15. Catalog not-null constraints

  16. parallel_schedule: add comment on event_trigger test dependency

  17. Revert "Catalog NOT NULL constraints" and fallout

  18. Adjust contrib/sepgsql regression test expected outputs.

  19. Fix table name clash in recently introduced test

  20. Catalog NOT NULL constraints

  21. Change the rules for inherited CHECK constraints to be essentially the same

On 28.02.23 20:15, Alvaro Herrera wrote:
> So I reworked this to use a new contype value for the NOT NULL
> pg_constraint rows; I attach it here.  I think it's fairly clean.
> 
> 0001 is just a trivial change that seemed obvious as soon as I ran into
> the problem.

This looks harmless enough, but I wonder what the reason for it is. 
What command can cause this error (no test case?)?  Is there ever a 
confusion about what table is in play?

> 0002 is the most interesting part.

Where did this syntax come from:

--- a/doc/src/sgml/ref/create_table.sgml
+++ b/doc/src/sgml/ref/create_table.sgml
@@ -77,6 +77,7 @@ CREATE [ [ GLOBAL | LOCAL ] { TEMPORARY | TEMP } | 
UNLOGGED ] TABLE [ IF NOT EXI

  [ CONSTRAINT <replaceable 
class="parameter">constraint_name</replaceable> ]
  { CHECK ( <replaceable class="parameter">expression</replaceable> ) [ 
NO INHERIT ] |
+  NOT NULL <replaceable class="parameter">column_name</replaceable> |
    UNIQUE [ NULLS [ NOT ] DISTINCT ] ( <replaceable 
class="parameter">column_name</replaceable> [, ... ] ) <replaceable 
class="parameter">in>
    PRIMARY KEY ( <replaceable 
class="parameter">column_name</replaceable> [, ... ] ) <replaceable 
class="parameter">index_parameters</replac>
    EXCLUDE [ USING <replaceable 
class="parameter">index_method</replaceable> ] ( <replaceable 
class="parameter">exclude_element</replaceable>

I don't see that in the standard.

If we need it for something, we should at least document that it's an 
extension.

The test tables in foreign_key.sql are declared with columns like

     id bigint NOT NULL PRIMARY KEY,

which is a bit weird and causes expected output diffs in your patch.  Is 
that interesting for this patch?  Otherwise I suggest dropping the NOT 
NULL from those table definitions to avoid these extra diffs.

> 0003:
> Since nobody liked the idea of listing the constraints in psql \d's
> footer, I changed \d+ so that the "not null" column shows the name of
> the constraint if there is one, or the string "(primary key)" if the
> attnotnull marking for the column comes from the primary key.  The new
> column is going to be quite wide in some cases; if we want to hide it
> further, we could add the mythical \d++ and have *that* list the
> constraint name, keeping \d+ as current.

I think my rough preference here would be to leave the existing output 
style (column header "Nullable", content "not null") alone and display 
the constraint name somewhere in the footer optionally.  In practice, 
the name of the constraint is rarely needed.

I do like the idea of mentioning primary key-ness inside the table somehow.

As you wrote elsewhere, we can leave this patch alone for now.