Re: PoC: Make it possible to disallow WHERE-less UPDATE and DELETE

On Mon, Sep 19, 2016 at 4:02 PM, David Fetter <david@fetter.org> wrote:
>
> [training_wheels_004.patch]

openjade:filelist.sgml:144:16:E: character "_" invalid: only parameter
literal, "CDATA", "ENDTAG", "MD", "MS", "PI", "PUBLIC", "SDATA",
"STARTTAG", "SYSTEM" and parameter separators allowed
openjade:contrib.sgml:138:2:W: cannot generate system identifier for
general entity "require"

The documentation doesn't build here, I think because require_where is
not an acceptable entity name.  It works for me if I change the
underscore to a minus in various places.  That fixes these errors:

+ <para>
+  Here is an example showing how to set up a database cluster with
+  <literal>require_where</literal>.
+<screen>
+$ psql -U postgres
+# SHOW shared_preload_libraries; /* Make sure not to clobber
something by accident */
+
+If you found something,
+# ALTER SYSTEM SET
shared_preload_libraries='the,stuff,you,found,require_where';
+
+Otherwise,
+# ALTER SYSTEM SET shared_preload_libraries='require_where';
+
+Then restart <productname>PostgreSQL</productname>
+</screen>
+ </para>

Could use a full stop (period) on the end of that sentence.  Also it
shouldn't be inside the "screen" tags.  Maybe "If you found
something," and "Otherwise," shouldn't be either, or should somehow be
marked up so as not to appear to be text from the session.

postgres=# delete from foo;
ERROR:  DELETE requires a WHERE clause
HINT:  To delete all rows, use "WHERE true" or similar.

Maybe one of those messages could use some indication of where this is
coming from, for surprised users encountering this non-standard
behaviour for the first time?

FWIW I saw something similar enforced globally by the DBA team at a
large company with many database users.  I think experienced users
probably initially felt mollycoddled when they first encountered the
error but I'm sure that some were secretly glad of its existence from
time to time...  I think it's a useful feature for users who want it,
and a nice little demonstration of how extensible Postgres is.

--
Thomas Munro
http://www.enterprisedb.com

--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers