From: | andrewsn(at)pgfoundry(dot)org (User Andrewsn) |
---|---|
To: | pgsql-committers(at)postgresql(dot)org |
Subject: | hstore-new - hstore-new: Deal with NULL issues in populate_record. |
Date: | 2009-09-14 03:30:29 |
Message-ID: | 20090914033029.6C0E31071FC8@pgfoundry.org |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-committers |
Log Message:
-----------
Deal with NULL issues in populate_record. We cope with the various issues
as follows:
- if we're passed in a null datum, we ensure that every field in the
returned heaptuple has been through its input function, even the null
values for fields that weren't set in the hstore
- if we're passed in a non-null datum heaptupleheader, we assume it's the
caller's job to have done any validation, domain checks, whatever, and
so we only call the input function on columns that we're overriding
with data from the hstore (even if the overriding value is a null)
Add appropriate regression tests.
Modified Files:
--------------
hstore-new:
hstore_io.c (r1.5 -> r1.6)
(http://cvs.pgfoundry.org/cgi-bin/cvsweb.cgi/hstore-new/hstore-new/hstore_io.c?r1=1.5&r2=1.6)
hstore-new/expected:
hstore.out (r1.5 -> r1.6)
(http://cvs.pgfoundry.org/cgi-bin/cvsweb.cgi/hstore-new/hstore-new/expected/hstore.out?r1=1.5&r2=1.6)
hstore-new/sql:
hstore.sql (r1.5 -> r1.6)
(http://cvs.pgfoundry.org/cgi-bin/cvsweb.cgi/hstore-new/hstore-new/sql/hstore.sql?r1=1.5&r2=1.6)
From | Date | Subject | |
---|---|---|---|
Next Message | Peter Eisentraut | 2009-09-14 13:23:48 | pgsql: Print builds don't actually depend on html target (anymore). |
Previous Message | User Andrewsn | 2009-09-14 00:32:56 | hstore-new - hstore-new: revisit previous fix; unconditionally set |