Re: [GSOC] questions about idea "rewrite pg_dump as library"

Michael Paquier <michael.paquier@gmail.com> writes:
> I recall discussions about reverse engineering of a parsed query tree in
> the event trigger threads but nothing has been committed I think. Also, you

Yes. The name used in there was "Normalized Command String".

> need to consider that implementing such reverse engineering mechanism in
> core might not be a good thing for new features and maintenance, as it
> would mean that it is necessary to change those APIs consistently with what
> is added on the parsing side.

The goal is to retain the same API, which is quite simple:
 function get_command_string(Node *parsetree) returns text

At the SQL level, the Node * is of datatype "internal" and you can't
forge it, you need to be given it in some ways. In the Event Trigger
case we though of a TG_PARSETREE magic variable, or maybe another
function get_current_parsetree() that only work when called from an
event trigger.

The other part of the API of course is how to represent the data, and as
we're talking about a Normalized Command String, there's no choice but
issuing a valid SQL command string that the server would know how to
execute and which would have the same side effects.

So of course a 9.3 and a 9.4 server equiped with that hypothetical
function would behave differently when the syntax did change. And that's
exactly why I think it the best choice here is to have that code
embedded and maintained in core.

Regards,
-- 
Dimitri Fontaine
http://2ndQuadrant.fr     PostgreSQL : Expertise, Formation et Support