Re: Column COMMENTs in CREATE TABLE? - Mailing list pgsql-hackers

From David G. Johnston
Subject Re: Column COMMENTs in CREATE TABLE?
Date
Msg-id CAKFQuwa0LED7S=OarPOAkNcysyYosXebkyQDTTX4LHcK36SAUQ@mail.gmail.com
Whole thread Raw
In response to Re: Column COMMENTs in CREATE TABLE?  (Fabrízio de Royes Mello <fabriziomello@gmail.com>)
List pgsql-hackers
On Sat, Jul 2, 2016 at 8:31 PM, Fabrízio de Royes Mello <fabriziomello@gmail.com> wrote:


Em sábado, 2 de julho de 2016, David G. Johnston <david.g.johnston@gmail.com> escreveu:
On Sat, Jul 2, 2016 at 12:55 PM, Marko Tiikkaja <marko@joh.to> wrote:

What I would prefer is something like this:

CREATE TABLE foo(
  f1 int NOT NULL COMMENT
    'the first field',
  f2 int NOT NULL COMMENT
    'the second field',
...
);

which would ensure the comments are both next to the field definition they're documenting and that they make it all the way to the database. I looked into the biggest products, and MySQL supports this syntax.  I couldn't find any similar syntax in any other product.


​+1 for the idea - though restricting it to columns would not be ideal.


CREATE TABLE name 
COMMENT IS 
'Table Comment Here'
(
col1 serial COMMENT IS 'Place comment here'
)​;


And what about the other CREATE statements? IMHO if we follow this path then we should add COMMENT to all CREATE statements and perhaps also to ALTER. Of course in a set of small patches to make the reviewers life easier.


​I should have made it clear I didn't expect TABLE to be the only object but rather was using it as an example of how we could/should do this generally for top-level objects (e.g., table) and sub-objects (e.g., column)​.

David J.

pgsql-hackers by date:

Previous
From: David Steele
Date:
Subject: Re: Password identifiers, protocol aging and SCRAM protocol
Next
From: Peter Geoghegan
Date:
Subject: Re: Bug in batch tuplesort memory CLUSTER case (9.6 only)