Justin Clift <jc at telstra.net> writes:
> Christopher Browne wrote:
>> Justin Clift <jc at telstra.net> writes:
> <snip>
>> Hmm.  I'd be game to create a descriptive table for that, and have
>> sl_event point to that as a "foreign key" validation...
>
> No idea how that would change things.  Those values are presently
> pretty much hard coded into the slony code though.
>
> But, I'm not a C expert so couldn't really say.  :)

Hey, this is not voodoo magic stuff.

Practically all of the functionality in slonik involves invoking
pl/pgsql functions, and, to be pedantic, this particular bit of
functionality does fit into what would get handled in pl/pgsql.

You'll find that pretty much any of the error messages that are raised
by slonik scripts that aren't about syntax (and therefore caught by
yacc/bison) are generated in one of two ways:

  1.  pl/pgsql code may check for error conditions, and raise errors;

  2.  You'll see pretty standard sorts of SQL exceptions involving
      non-unique indices, missing foreign keys, and such.

"C expert" isn't a requirement in any of that.  I'll see about putting
an "autodoc" dump of the schema and functions into CVS soon; I think
that should be helpful in dispelling some of this magic...
-- 
"cbbrowne","@","ca.afilias.info"
<http://dev6.int.libertyrms.com/>
Christopher Browne
(416) 673-4124 (land)