Re: improved DefElem list processing

On 8/5/16 11:25 AM, Peter Eisentraut wrote:
> On 8/4/16 2:21 PM, Tom Lane wrote:
>> Forgot to mention: seems like you should have added a location
>> argument to makeDefElem.
>
> I was hesitating to do that lest it break extensions or something, but I
> guess we break bigger things than that all the time.  I'll change it.

In order not to work on two patches that directly conflict with each
other, I have proceeded with the location patch and postponed the
duplicate checking patch.

Attached is a biggish patch to review.  It adds location information to
all places DefElems are created in the parser and then adds errposition
information in a lot of places, but surely not all of them.  That can be
improved over time.

I'm not happy that utils/acl.h has prototypes for aclchk.c, because
acl.h is included all over the place.  Perhaps I should make a
src/include/catalog/aclchk.c to clean that up.

Here are some example commands to try for getting suitable error messages:

create collation foo (foo = bar, bar = foo);
copy test from stdin (null 'x', null 'x');
create function foo (a int, b int) returns int as $$ select a+b $$
language sql language sql;
create function foo (a int, b int) returns int as $$ select a+b $$
language sql volatile stable;
create function foo (a int, b int) returns int as $$ select a+b $$
language sql with (foo = bar);
create sequence foo minvalue 1 minvalue 2;
create type foo (foo = bar);
create user foo createdb nocreatedb;
explain (foo, bar) select 1;

--
Peter Eisentraut              http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services