pg_catalog.int does not exist about piggly

[kputnam]

This could be a PostgreSQL version issue. Do the examples also fail with examples/run-tests?

[lcoldiron]

I'll check when I get back to my desk. I think I have a fix that will work. I can send you the code since I can't upload it at work.

Sent from my HTC

----- Reply message -----
From: "kputnam" [email protected]
Date: Fri, Aug 12, 2011 12:13 pm
Subject: [piggly] pg_catalog.int does not exist (#19)
To: [email protected]

This could be a PostgreSQL version issue. Do the examples also fail with examples/run-tests?

Reply to this email directly or view it on GitHub:
#19 (comment)

[lcoldiron]

I ran the example/run-tests and don't see the issue but I think it might be because none of the example procs have a return type of int.

CREATE OR REPLACE FUNCTION public.my_func()
RETURNS "pg_catalog"."int" AS
$BODY$
BEGIN
END;
$BODY$
LANGUAGE plpgsql VOLATILE SECURITY DEFINER;

I tried the following in Postgres 8.3 and 8.4 and it doesn't seem to work.

My fix for this is a little bit of a hack but in lib/piggly/dumper/procedure.rb line:232
Before: q(hash["typens"], shorten(hash["type"])),
After: hash["typens"] == "pg_catalog" ? q(hash["type"]) : q(hash["typens"], shorten(hash["type"])),

This seems to work for the situation where you need to schema qualify for row types and when you don't for primitive types.

[kputnam]

Yeah, that's basically what I had in mind for a fix. Might as well get rid of references to the "pg_catalog" schema altogether... the initialize method for QualifiedName should probably be

def initialize(name, *names)
  @names =
    if name == "pg_catalog" and not names.empty?
      names
    else
      [name, *names]
    end
end

Feel free to commit this and push when you get a chance.

[kputnam]

Still noticing a problem -- QualifiedName#quote doesn't suppress "pg_catalog".

[kputnam]

Finally took the 30 minutes to sit down and figure this out. Should be OK now!